Lead Researcher of Computer Vision & Lead of Overseas AI Development Group
R&D Department, Chowagiken Corp., Japan
Duration: 01 May 2023 to now
Follow me:
Dr. Md Rashedul Islam
Lead Researcher of Computer Vision
and Lead of Overseas AI Development Group
R&D Department, Chowagiken Corp., Japan
[Former] Senior Engineer, R&D department,Nikon-Exvision Corporation, Tokyo, Japan
[Former] Visiting Researcher (postdoc)[Visit ]
School of Computer Science and Engineering, University of Aizu, Japan
[Former] Associate Professor [Visit ]
Dept. of CSE, University of Asia Pacific, Dhaka, Bangladesh
Supervisor [Visit ]
Computer Vision & Pattern Recognition Lab, University of Asia Pacific
PhD, Embedded System Lab, University of Ulsan, South Korea
M.Sc., University of Boras, Sweden
B.Sc. (Hons), University of Rajshahi, Bangladesh
Senior member, IEEE
Download CV
Short Biography
Dr. Md. Rashedul Islam received a B.Sc. degree in computer science and engineering from the University Rajshahi, Rajshahi, Bangladesh, in 2006, an M.Sc. degree in informatics from the Högskolan i Borås (University of Boras), Boras, Sweden, in 2011, and a Ph.D. degree in electrical, electronic, and computer engineering at the University of Ulsan, Ulsan, South Korea, in 2016. He is currently working as a Lead Researcher of Computer Vision and Lead of Overseas AI Development Group, Chowagiken Corp., Japan. Previous, he worked as a Senior Engineer, R&D department, Nikon-Exvision Corporation, Tokyo, Japan; Visiting researcher (postdoc) in the School of Computer Science and Engineering, University of Aizu, Japan; Associate Professor in the Department of Computer Science and Engineering, University of Asia Pacific (UAP), Dhaka, Bangladesh; Graduate research assistant in the Embedded system lab, University of Ulsan, South Korea; Assistant professor in the Department of Computer Science and Engineering, University of Asia Pacific (UAP), Dhaka, Bangladesh; and Lecturer in the Department of Computer Science and Engineering, Leading University, Sylhet, Bangladesh. His research areas are Computer Vision, Machine learning, deep learning, AI, Signal & Image processing, HCI, health informatics, Bearing fault diagnosis, and others. He is a reviewer of several journals, like, IEEE Transactions on industrial electronics, IEEE Access, Applied Science, Multimedia tools and application, Cluster Computing, Shock and Vibration, Journal of Information Processing Systems, and others. He is also a PC member of several international conferences. Also, he has a good experience in professional IT system analysis and development. He is a Senior member of the IEEE Computer Society and IEEE Computational Intelligence Society.
He has also served as: a) Secretary, organizing committee, 19th International Conference on Computer and Information Technology 2017 (ICCIT2017), b) Organizing Chair, Organizing committee, ACM-ICPC Dhaka Regional Site 2017, c) Head, Self-Assessment Committee (SAC) of Dept. of CSE Under IQAC, University of ASIA PACIFIC, d) Coordinator, MCSE Program, Dept. of CSE, University of ASIA PACIFIC, e) Convener of Software and Hardware Club, Dept. of CSE, University of ASIA PACIFIC. f) Coordinator, Admission Committee, Dept. of CSE, University of ASIA PACIFIC, g) Treasurer, Bangladesh Advanced Computing Society.
Education & Training
-
Ph.D. Aug 2016
Ph.D. in Computer Engineering
Embedded System Lab, Department of Electrical, Electronics and Computer Engineering,
University of Ulsan, Ulsan, South Korea
Thesis: Discriminant Fault Feature Selection and Reliable Online Bearing Fault Diagnosis System using Signal Processing and Machine Learning Techniques -
M.Sc.2011
M.Sc. in Informatics
University of Borås (Högskolan i Borås), Sweden.
Thesis: Evaluations of the parallel extensions in .NET 4.0 -
B.Sc2006
B.Sc. in Computer Science and Engineering
Department of Computer Science and Engineering,
University of Rajshahi, Bangladesh
Thesis: Speaker Identification System Using Eigenface Classification Engine Area: Speech Signal Processing
-
HSC 2000
Higher secondary certificate
Major in Science, Govt. A. H. College, Bogra under Rajshahi board, Bangladesh.
-
SSC1998
Secondary school certificate
Major in Science, Gabtali Pilot high school, Bogra, under Rajshahi board, Bangladesh.
Work Experiences
Academic and research experience
-
Present 2023
-
2023 2020
Senior Engineer, R&D department, Industry Solutions Division
Nikon-Exvision Corporation, Tokyo, Japan
Duration: 01 January 2020 to 30 April 2023 -
2019 2018
Visiting Researcher (Postdoc)
School of Computer Science & Engineering, University of Aizu, Fukushima, Japan
Duration: 01 April 2018 to December 2019 -
2018 2016
Associate Professor
Dept. of Computer Science & Engineering, University of Asia Pacific, Dhaka, Bangladesh
Duration: 16-10-2016 to 30-03-2018 -
2016 2011
Assistant Professor
Dept. of Computer Science & Engineering, University of Asia Pacific, Dhaka, Bangladesh
Duration: 09-10-2011 to 15-10-2016 -
2016 2013
Graduate Research Assistant
Embedded System Lab, University of Ulsan, South Korea. (http://eucs.ulsan.ac.kr)
Duration: 01-09-2013 to 22-08-2016 -
2011 2009
Senior Lecturer
Dept. of Computer Science & Engineering, Leading University, Sylhet, Bangladesh
Duration: 17-09-2009 to 01-10-2011 -
2009 2007
Lecturer
Dept. of Computer Science & Engineering, Leading University, Sylhet, Bangladesh
Duration: 17-09-2007 to 16-09-2009
Professional experience
-
Present 2016
Software Engineering Consultant
Orbit Itech Ltd., Birminghum, UK
Duration: 01-Feb-2016 to till date -
2012 2011
IT consultant
Arrowsoft, Sylhet, Bangladesh
Duration: Jan 2011 to June 2012 -
2007 2007
Chief Technical Officer
Firm: Orbit Solutions Ltd (Complete IT Solution Firm)
Niloy-60(2nd Floor), Dorga gate, Sylhet.
Duration: 01-03-2007 To 15-09-2007 -
2007 2006
Senior Developer
Firm: Orbit Solutions Ltd (Complete IT Solution Firm),
Niloy-60(2nd Floor), Dorga gate, Sylhet.
Duration: 01-09-2006 To 28-02-2007
Other Experiences
-
2018 2017
Head
Self-Assessment Committee (SAC) of Dept. of CSE Under IQAC
-
2017
Secretary
Organizing committee,19th International Conference on Computer and Information Technology 2017 (ICCIT2017).
-
2017
Chair
Organizing committee, ACM-ICPC Dhaka Regional Site 2017
-
2017 2016
Convener
Software and Hardware Club, Dept. of CSE, University of Asia Pacific
-
2017 2016
Coordinator
Admission Committee, Dept. of CSE, University of Asia Pacific
-
2018 2017
Treasurer
Bangladesh Advanced Computing Society.
-
2013 2011
Coordinator
Master of Science in Computer Science and Engineering (MCSE), University of Asia, Pacific
-
2013 2011
Convener
Research and Publication Unit, Dept of CSE, University of Asia Pacific
-
2009
Head (Acting)
Dept. of CSE, Leading University
-
2009 2007
Adviser
Computer Club, Dept. of CSE, Leading University, Sylhet
Researches
Research Interests
- Artificial Intelligence
- Machine learning & Deep learning
- Coputer Vision
- Signal & Image processing
- Feature selection & optimization
- Data Science
- Industrial anomaly detection
- Human activity recognition
- Health informatics & disease diagnosis
- Human Computer Interaction
- Robitic applications
- Fault detection and diagnosis of induction motor
Research Projects
-
--
-
--
-
--
-
--
-
--
-
--
-
--
-
--
-
--
Thesis Supervisions
-
PhD
2019~Human Activity Recognition and Elderly People Fall Prediction using Machine Learning and Signal Processing techniques
Department of Computer Science and Engineering, Bangladesh University of Professionals -
PhD (Co-supervision)
2021-2024Cloud-based Autonomous Robotic Systems for Enhanced Mobility and Accessibility
School of Computer Science and Engineering, University of Aizu, Japan -
PhD (Co-supervision)
2018-2021Hand Gesture and Handwriting Based HCI and Health Informatics System Using Machine Learning Techniques
School of Computer Science and Engineering, University of Aizu, Japan -
B.Sc. thesis
Fall 2018Enhanced denoising of sEMG signal for efficient human action recognition
School of Computer Science and Engineering, University of Aizu -
B.Sc. Thesis
Fall 2018Brain computer interaction based of EEG signal analysis
Dept. of Computer Science and Engineering, University of Asia Pacific -
B.Sc. Thesis
Fall 2018Outlier detection based leaf disease detection and diagnosis using image processing techniques
Dept. of Computer Science and Engineering, University of Asia Pacific -
B.Sc. Thesis
Fall 2018Parkinson"s disease detection based on hand writing analysis
Dept. of Computer Science and Engineering, University of Asia Pacific -
B.Sc. Thesis
Spring 2018Real time Human computer interaction based on dynamic hand gesture
Dept. of Computer Science and Engineering, University of Asia Pacific -
M.Sc Thesis
Fall 2017PCA and ICA Based Feature Extraction for Cardiac Arrhythmia Disease Diagnosis.
Dept. of Computer Science and Engineering, University of Asia Pacific -
M.Sc. Thesis
Fall 2017Enhanced Sleep Disorder Detection Method Using DWT and Feature Selection Technique
Dept. of Computer Science and Engineering, University of Asia Pacific -
M.Sc Thesis
Fall 2017Feature Extraction from EEG signal Using EMD for Sleep Disorder Detection
Dept. of Computer Science and Engineering, University of Asia Pacific -
B.Sc Thesis
Fall 2017Video Based Smoke Detection Using Smoke Growth Analysis for Fire Alarm System
Dept. of Computer Science and Engineering, University of Asia Pacific -
B.Sc Thesis
Fall 2017A Fault Diagnosis Model of Bearing using Convolutional Neural Network
Dept. of Computer Science and Engineering, University of Asia Pacific -
B.Sc Thesis
Spring 2017Reduction of Gesture Feature Dimension for Improving the Hand Gesture Recognition Performance of Numerical Sign Language
Dept. of Computer Science and Engineering, University of Asia Pacific -
M.Sc Thesis
Spring 2013Performance evaluation on Data partitioning and task partitioning in parallelism
Dept. of Computer Science and Engineering, University of Asia Pacific -
B.Sc Thesis
Spring 2013Memory Efficient adaptive Distributed database management system using cost function
Dept. of Computer Science and Engineering, University of Asia Pacific
Total:47
2025(4) 2024(5) 2023(2) 2022(4) 2021(4) 2020(5) 2019(4) 2018(1) 2017(3) 2016(6) 2015(1) 2014(3) 2012(1) 2011(4)
Total:44
2024(1) 2023(1) 2022(1) 2021(4) 2020(4) 2019(5) 2018(7) 2017(7) 2016(1) 2015(2) 2014(6) 2009(3) 2006(2)
Total:3
2014(2) 2013(1)
Total:4
2017(1) 2016(2) 2015(1)
Total:1
2025(1)
Publications
International Journal
Total:47
2025(4) 2024(5) 2023(2) 2022(4) 2021(4) 2020(5) 2019(4) 2018(1) 2017(3) 2016(6) 2015(1) 2014(3) 2012(1) 2011(4)
International Conference
Total:44
2024(1) 2023(1) 2022(1) 2021(4) 2020(4) 2019(5) 2018(7) 2017(7) 2016(1) 2015(2) 2014(6) 2009(3) 2006(2)
Domestic Conference
Total:3
2014(2) 2013(1)
Book/Book Chapters
Total:4
2017(1) 2016(2) 2015(1)
Book Chapters
Total:1
2025(1)
List of Publication (pdf)
Please select type to see specific publications
Filter by type:
Sort by year:
International Journal On June-2025
Hybrid Contrastive Learning with Attention-Based Neural Networks for Robust Fraud Detection in Digital Payment Systems
IEEE Open Journal of the Computer Society
SCI | Q1 | Impact Factor: 8.2
Abstract
Fraud detection in digital payment systems is a critical challenge due to the growing complexity of transaction patterns and the inherent class imbalance in datasets. This paper proposes a novel Hybrid Contrastive Learning framework integrating Siamese Networks with Attention-Based Neural Networks to effectively distinguish fraudulent from legitimate transactions. The proposed model achieves state-of-the-art results, surpassing 10 recent methods in key metrics, with a recall of 95.42%, precision of 97.35%, and ROC-AUC of 98.78% on the Credit Card Fraud Detection dataset. Cross-dataset evaluations using a simulated transaction dataset demonstrate consistent generalization, achieving a recall of 95.12% and ROC-AUC of 98.60%. An ablation study underscores the impact of attention mechanisms and contrastive learning, with the combined approach improving F1-score by up to 2.64%. Additionally, SHAP-based analysis reveals the importance of key features such as transaction amount and PCA-derived components in model decisions, enhancing explainability. The results establish the proposed framework as a robust, interpretable, and scalable solution for fraud prevention in digital payment systems.
Book Chapters On March-2025
Advancements in AI-Based Anomaly Detection for Smart Manufacturing
Springer Series in Advanced Manufacturing
Abstract
In the Industry 4.0 era, smart manufacturing has become an integral part of today’s modern industrial space with the integration of advanced technologies such as the Artificial intelligence (AI), big data analytics, and Internet of Things (IoT). This chapter provides a comprehensive review of recent advancements in AI-based anomaly detection techniques tailored for smart manufacturing contexts. Different machine learning (ML) and deep learning (DL) techniques, including supervised learning, unsupervised learning, semi-supervised learning, reinforcement learning, and hybrid learning, explore the applicability of anomaly detection for various types of anomalies, address challenges of class imbalance and the scarcity of labeled data, and adapt to various industrial environments. Leveraging ML and DL methodologies and the large number of multipurpose anomaly benchmark datasets, AI has demonstrated superior accuracy and efficiency in identifying anomalies crucial for maintaining operational efficiency, ensuring product quality, and safeguarding human workers and machinery. Furthermore, the chapter looks into future trends and potential improvements in AI-based anomaly detection, considering the challenges for integrating the AI systems into existing infrastructure. By addressing these challenges, AI-based anomaly detection can significantly enhance the reliability and productivity of smart manufacturing systems.
International Journal On March-2025
A Dual Output Temporal Convolutional Network with Attention Architecture for Stock Price Prediction and Risk Assessment
IEEE Access
SCIE | Impact Factor: 3.476
Abstract
In this paper, we propose a novel deep learning model that integrates a Temporal Convolutional Network (TCN) with an Attention mechanism to predict stock prices and assess risk for MasterCard (MA) and Visa (V). The model is designed with a dual output to forecast future stock prices (Open, Close, High, Low) while simultaneously predicting risk metrics, including volatility and the Sharpe Ratio. By utilizing dilated convolutions, the TCN captures both short-term and long-term dependencies in the stock price data, and the attention layer focuses on critical time steps for enhanced predictive accuracy. On a dataset covering over 15 years (2008-2024), the TCN with attention model achieved a mean absolute error (MAE) of 1.23 for MasterCard and 1.45 for Visa, outperforming LSTM and ARIMA baselines. Additionally, the model demonstrated strong performance in predicting risk metrics, with an MAE of 0.012 for volatility and 0.065 for the Sharpe Ratio. Backtesting on unseen data further validated the model’s robustness, with a backtest MAE of 1.25 for MasterCard and 1.50 for Visa. This dual-output architecture provides an accurate and interpretable solution for both stock price forecasting and risk assessment, offering a valuable tool for investors and risk managers.
International Journal On February-2025
A Hybrid Temporal Convolutional Network and Transformer Model for Accurate and Scalable Sales Forecasting
IEEE Open Journal of the Computer Society
SCI | Q1 | Impact Factor: 5.7
Abstract
Accurate product sales forecasting is critical for inventory management, pricing strategies, and supply chain optimization in the retail industry. This article proposes a novel deep learning architecture that integrates Temporal Convolutional Networks (TCNs) with Transformer-based attention mechanisms to capture both short-term and long-term dependencies in time-series sales data. Utilizing the Favorita Grocery Sales Forecasting dataset, our hybrid TCN Transformer model demonstrates superior performance over existing models by incorporating external factors such as holidays, promotions, oil prices, and transaction data. The model achieves state-of-the-art results with a Mean Absolute Error (MAE) of 2.01, Root Mean Squared Error (RMSE) of 2.81, and a Weighted Mean Absolute Percentage Error (wMAPE) of 4.22%, significantly outperforming other leading models such as LSTM, GRU, and TFT. Extensive cross-validation confirms the robustness of our model, achieving consistently high performance across multiple folds.
International Journal On January-2025
Deep Learning Model Based Prediction of Vehicle CO2 Emissions with eXplainable AI Integration for Sustainable Environment
Nature scientific reports
SCIE | Q1 | Impact Factor: 3.8
Abstract
The transportation industry contributes significantly to climate change through carbon dioxide (CO2) emissions, intensifying global warming and leading to more frequent and severe weather phenomena such as flooding, drought, heat waves, glacier melting, and rising sea levels. This study proposes a comprehensive approach for predicting CO2 emissions from vehicles using deep learning techniques enhanced by eXplainable Artificial Intelligence (XAI) methods. Utilizing a dataset from the Canadian government’s official open data portal, we explored the impact of various vehicle attributes on CO2 emissions. Our analysis reveals that not only do high-performance engines emit more pollutants, but fuel consumption under both city and highway conditions also contributes significantly to higher emissions. We identified skewed distributions in the number of vehicles produced by different manufacturers and trends in fuel consumption across fuel types. This study used deep learning techniques to construct a CO2 emission prediction model, specifically a light multilayer perceptron (MLP) architecture called CarbonMLP. The proposed model was optimized by hyperparameter tuning and achieved excellent performance metrics, such as a high R-squared value of 0.9938 and a low Mean Squared Error (MSE) of 0.0002. This study employs XAI approaches, particularly SHapley Additive exPlanations (SHAP), to improve the model interpretation ability and provide information about the importance of features. The findings of this study show that the proposed methodology accurately predicts CO2 emissions from vehicles. Additionally, the analysis suggests areas for further research, such as increasing the dataset, integrating additional pollutants, improving interpretability, and investigating real-world applications. Overall, this study contributes to the design of effective strategies for reducing vehicle CO2 emissions and promoting environmental sustainability.
International Journal On December-2024
Revolutionizing Rose Grading: Real-Time Detection and Accurate Assessment with YOLOv8 and Deep Learning Models
Springer Nature (SR) Computer Science
SCIE | Q2 | Impact Factor: 3.1
Abstract
Yield estimation and identifying the growth stages of roses greatly depend on the automatic detection of roses in orchards. YOLO is well known for its high accuracy and real-time performance among the deep learning-based object detection techniques. In our study, we proposed a model based on YOLOv8 and DNN that uses detection, localization, and classification to grade rose quality. YOLOv8 was used to localize the roses with bounding boxes. The regions of interest are detected by YOLOv8 and cropped from the source images by the bounding boxes and fed into deep learning model for classifying roses according to growth stages, thus, the freshness level is ultimately graded. We used YOLOv8 for detection and DNN for classification. We proposed two different deep learning models for the classification task. The first was a lightweight CNN model which achieved a faster speed in grading roses, while the second was a fine-tuned MobileNet model which achieved a higher accuracy. We have also collected a new dataset that contained variety, complex backgrounds, overlapping roses etc. Numerous deep learning models were exploited in this work, including ResNet50, VGG19, MobileNet, NASNetMobile for comparison purpose. Comparing with these models, proposed CNN and fine-tuned MobileNet had simple and light-weight structure, better accuracy and our method could fulfill the real-time requirements. In our proposed method, YOLOv8 was used to detect roses. Then after applying fine-tuning, fine-tuned MobileNet achieved remarkable accuracy in classifying roses. The study may give technological references for developing rose harvesting robots and estimating orchard yields.
International Conference On November-2024
Bayesian and Genetic Optimization for Human Activity Recognition with CNN and LSTM
International Conference on Intelligent and Innovative Practices in Engineering & Management IIPEM 2024
Abstract
Recent advancement of wearable sensor technology causes an increase use on human activity recognition (HAR) due to its superior performance and portability. HAR is applied on potential applications such as ambient assisted living, home automation, healthcare, games, monitoring, and inquiring activities. This study focuses on predicting human activities from the PAMAP2 dataset utilizing Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) models. Both models are optimized using Bayesian Optimization (BO) and Genetic Algorithm (GA) techniques precise classification. The research started with a comprehensive Exploratory Data Analysis (EDA) to discover the fundamental construction of activity data. CNN and LSTM models are then implemented and optimized for hyperparameter tuning, enhancing predictive performance. The proposed study is assessed on various metrics, with accuracy, which ranged from 93.8% to 95.35%. Furthermore, a comparative analysis is conducted to identify the most effective model for human activity prediction, highlighting the strengths and weaknesses of each approach.
International Journal On October-2024
A hybrid transformer and attention based recurrent neural network for robust and interpretable sentiment analysis of tweets
Nature scientific reports
SCIE | Q1 | Impact Factor: 3.8
Abstract
Sentiment analysis is a pivotal tool in understanding public opinion, consumer behavior, and social trends, underpinning applications ranging from market research to political analysis. However, existing sentiment analysis models frequently encounter challenges related to linguistic diversity, model generalizability, explainability, and limited availability of labeled datasets. To address these shortcomings, we propose the Transformer and Attention-based Bidirectional LSTM for Sentiment Analysis (TRABSA) model, a novel hybrid sentiment analysis framework that integrates transformer-based architecture, attention mechanism, and recurrent neural networks like BiLSTM. The TRABSA model leverages the powerful RoBERTa-based transformer model for initial feature extraction, capturing complex linguistic nuances from a vast corpus of tweets. This is followed by an attention mechanism that highlights the most informative parts of the text, enhancing the model’s focus on critical sentiment-bearing elements. Finally, the BiLSTM networks process these refined features, capturing temporal dependencies and improving the overall sentiment classification into positive, neutral, and negative classes. Leveraging the latest RoBERTa-based transformer model trained on a vast corpus of 124M tweets, our research bridges existing gaps in sentiment analysis benchmarks, ensuring state-of-the-art accuracy and relevance. Furthermore, we contribute to data diversity by augmenting existing datasets with 411,885 tweets from 32 English-speaking countries and 7,500 tweets from various US states. This study also compares six word-embedding techniques, identifying the most robust preprocessing and embedding methodologies crucial for accurate sentiment analysis and model performance. We meticulously label tweets into positive, neutral, and negative classes using three distinct lexicon-based approaches and select the best one, ensuring optimal sentiment analysis outcomes and model efficacy. Here, we demonstrate that the TRABSA model outperforms the current seven traditional machine learning models, four stacking models, and four hybrid deep learning models, yielding notable gain in accuracy (94%) and effectiveness with a macro average precision of 94%, recall of 93%, and F1-score of 94%. Our further evaluation involves two extended and four external datasets, demonstrating the model’s consistent superiority, robustness, and generalizability across diverse contexts and datasets. Finally, by conducting a thorough study with SHAP and LIME explainable visualization approaches, we offer insights into the interpretability of the TRABSA model, improving comprehension and confidence in the model’s predictions. Our study results make it easier to analyze how citizens respond to resources and events during pandemics since they are integrated into a decision-support system. Applications of this system provide essential assistance for efficient pandemic management, such as resource planning, crowd control, policy formation, vaccination tactics, and quick reaction programs.
International Journal On July-2024
A Robust Deep Feature Extraction Method for Human Activity Recognition Using a Wavelet Based Spectral Visualisation Technique
Special Issue Human Activity Recognition Using Sensors and Machine Learning: 2nd Edition, Sensors
SCIE | Q1 | Impact Factor: 3.4
Abstract
Human Activity Recognition (HAR), alongside Ambient Assisted Living (AAL), are integral components of smart homes, sports, surveillance, and investigation activities. To recognize daily activities, researchers are focusing on lightweight, cost-effective, wearable sensor-based technologies as traditional vision-based technologies lack elderly privacy, a fundamental right of every human. However, it is challenging to extract potential features from 1D multi-sensor data. Thus, this research focuses on extracting distinguishable patterns and deep features from spectral images by time-frequency-domain analysis of 1D multi-sensor data. Wearable sensor data, particularly accelerator and gyroscope data, act as input signals of different daily activities, and provide potential information using time-frequency analysis. This potential time series information is mapped into spectral images through a process called use of ’scalograms’, derived from the continuous wavelet transform. The deep activity features are extracted from the activity image using deep learning models such as CNN, MobileNetV3, ResNet, and GoogleNet and subsequently classified using a conventional classifier. To validate the proposed model, SisFall and PAMAP2 benchmark datasets are used. Based on the experimental results, this proposed model shows the optimal performance for activity recognition obtaining an accuracy of 98.4% for SisFall and 98.1% for PAMAP2, using Morlet as the mother wavelet with ResNet-101 and a softmax classifier, and outperforms state-of-the-art algorithms.
International Journal On July-2024
DCNN: Deep Convolutional Neural Network with XAI for Efficient Detection of Specific Language Impairment in Children
IEEE Access
SCIE | Q1 | Impact Factor: 3.476
Abstract
Assessing children for specific language impairment (SLI) or other communication impairments can be challenging for doctors due to the extensive battery of tests and examinations required. Artificial intelligence and computer-aided diagnostics have aided medical professionals in conducting rapid, reliable assessments of children’s neurodevelopmental conditions concerning language comprehension and output. Previous research has shown differences between the vocal characteristics of typically developing (TD) children and those with SLI. This study aims to develop a natural language processing (NLP) system that can identify children’s early impairments using specific conditions. Our dataset contains examples of disorders, and this study seeks to (1) demonstrate the effectiveness of several classifiers in this regard and (2) select the most effective model from the classifiers. We utilized various machine learning (ML), deep learning (DL), and transformer models to achieve our objective. Our deep convolutional neural network (DCNN) model yielded excellent results, outperforming the competition with an accuracy of 90.47%, making it the top-performing model overall. To increase the accuracy and credibility of our most likely output, we have incorporated explainable AI approaches like SHAP and LIME. These approaches aid in interpreting and explaining model predictions, considering the significance and sensitivity of the topic. Additionally, we believe that our work can contribute to developing more accessible, effective methods for diagnosing language impairments in young children.
International Journal On February-2024
Enhanced Robot Motion Block of A-Star Algorithm for Robotic Path Planning
Sensors
SCIE | Q1 | Impact Factor: 3.9
Abstract
An optimized robot path-planning algorithm is required for various aspects of robot movements in applications. The efficacy of the robot path-planning model is vulnerable to the number of search nodes, path cost, and time complexity. The conventional A-star (A*) algorithm outperforms other grid-based algorithms because of its heuristic approach. However, the performance of the conventional A* algorithm is suboptimal for the time, space, and number of search nodes, depending on the robot motion block (RMB). To address these challenges, this paper proposes an optimal RMB with an adaptive cost function to improve performance. The proposed adaptive cost function keeps track of the goal node and adaptively calculates the movement costs for quickly arriving at the goal node. Incorporating the adaptive cost function with a selected optimal RMB significantly reduces the searches of less impactful and redundant nodes, which improves the performance of the A* algorithm in terms of the number of search nodes and time complexity. To validate the performance and robustness of the proposed model, an extensive experiment was conducted. In the experiment, an open-source dataset featuring various types of grid maps was customized to incorporate the multiple map sizes and sets of source-to-destination nodes. According to the experiments, the proposed method demonstrated a remarkable improvement of 93.98% in the number of search nodes and 98.94% in time complexity compared to the conventional A* algorithm. The proposed model outperforms other state-of-the-art algorithms by keeping the path cost largely comparable. Additionally, an ROS experiment using a robot and lidar sensor data shows the improvement of the proposed method in a simulated laboratory environment.
International Journal On July-2023
G-BERT: An Efficient Method for Identifying Hate Speech in Bengali Texts on Social Media
IEEE Access
SCIE | Q1 | Impact Factor: 3.476
Abstract
International Journal On February-2023
An Unsupervised Writer Identification Based on Generating Clusterable Embeddings
Computer Systems Science and Engineering
SCIE | Q2 | Impact Factor: 4.397
Abstract
The writer identification system identifies individuals based on their handwriting is a frequent topic in biometric authentication and verification systems. Due to its importance, numerous studies have been conducted in various languages. Researchers have established several learning methods for writer identification including supervised and unsupervised learning. However, supervised methods require a large amount of annotation data, which is impossible in most scenarios. On the other hand, unsupervised writer identification methods may be limited and dependent on feature extraction that cannot provide the proper objectives to the architecture and be misinterpreted. This paper introduces an unsupervised writer identification system that analyzes the data and recognizes the writer based on the inter-feature relations of the data to resolve the uncertainty of the features. A pairwise architecture-based Autoembedder was applied to generate clusterable embeddings for handwritten text images. Furthermore, the trained baseline architecture generates the embedding of the data image, and the K-means algorithm is used to distinguish the embedding of individual writers. The proposed model utilized the IAM dataset for the experiment as it is inconsistent with contributions from the authors but is easily accessible for writer identification tasks. In addition, traditional evaluation metrics are used in the proposed model. Finally, the proposed model is compared with a few unsupervised models, and it outperformed the state-of-the-art deep convolutional architectures in recognizing writers based on unlabeled data.
International Conference On February-2023
A Machine Learning Based Analysis Between Climate Change and Human Health: A Correlational Study
2022 International Conference on Computer and Applications (ICCA)
Abstract
Climate change has huge impact in human health. Social and environmental determinants of health are affected by climate change. According to World Health Organization (WHO) states that the strokes, most heart diseases, cancers, diabetes, chronic kidney diseases are the top causes of death. Although there are direct factors behind these diseases, climate change could have an invisible role in the rise of these diseases. Researchers are using various technologies to find correlations between climate change and human health, particularly trying to find out which elements of the weather are more responsible. Although there are explicit reasons for the formation of these diseases. But few studies have been conducted on passive factors that have a hidden but serious effect on the formation of these diseases. In this regard, machine learning approach can help us to correlate between the features of climate and various human diseases. Following that, the study uses Pearson, Spearman and Phi-K algorithms to determine the possibilities of correlation between human health and climate change. The research states that Carbon Monoxide (CO) have 98% of correlation and carbon dioxide (CO 2 ) has 95% of correlation with cardiovascular disease (Ca).
International Journal On December-2022
A Comprehensive Survey on the Progress, Process, and Challenges of Lung Cancer Detection and Classification
Journal of Healthcare Engineering, Hindawi
SCIE | Q2 | Impact Factor: 3.822
Abstract
International Journal On September-2022
A Comprehensive Survey On The Process, Methods, Evaluation, and Challenges Of Feature Selection
IEEE Access, vol. 10
SCI | Q1 | Impact Factor: 3.476
Abstract
Feature selection is employed to reduce the feature dimensions and computational complexity by eliminating irrelevant and redundant features. A vast amount of increasing data and its processing generates many feature sets, which are reduced by the feature selection process to improve the performance in all types of classification, regression, clustering models. This study performs a detailed analysis of motivation and concentrates on the fundamental architecture of feature selection. This study aims to establish a structured formation related to popular methods such as filters, wrappers and, embedded into search strategies, evaluation criteria, and learning methods. Different methods organize a comparison of the benefits and drawbacks followed by multiple classification algorithms and standard validation measures. The diversity of applications in multiple domains such as data retrieval, prediction analysis, and medical, intrusion, and industrial applications is efficiently highlighted. This study focuses on some additional feature selection methods for handling big data. Nonetheless, new challenges have surfaced in the analysis of such data, which were also addressed in this study. Reflecting on commonly encountered challenges and clarifying how to obtain the absolute feature selection method are the significant components of this study.
International Conference On September-2022
Mental Disability Detection of Children Using Handwriting Analysis
Springer
Abstract
Handwriting analysis also known as graphology is a way of knowing particular traits and behavior respective of a person. So far studies related to graphology are mostly concerned with the prediction of behaviors and so this paper focuses on the use of handwriting for other mental conditions classification. With the increase in mental disorder found in children, having an automated system could be diagnosis friendly. And so, the proposed methodology is concerned with the development of a device for predicting the mental condition of children through automated handwriting analysis. In this paper, a pen tablet is used for the collection of handwriting data samples measuring features. A parameter selection process is used for top dominant parameters and features are extracted from selected parameters, which are act as the input to the model. For the classification of the model that is the mental condition of the child, we used six different algorithms as classifiers, and among these SVM and Decision tree holding the highest accuracy of 72.7%.
International Journal On March-2022
A Comprehensive Survey on the Detection, Classification, and Challenges of Neurological Disorders
Biology
SCIE | Q1 | Impact Factor: 5.079
Abstract
Neurological disorders (NDs) are becoming more common, posing a concern to pregnant women, parents, healthy infants, and children. Neurological disorders arise in a wide variety of forms, each with its own set of origins, complications, and results. In recent years, the intricacy of brain functionalities has received a better understanding due to neuroimaging modalities, such as magnetic resonance imaging (MRI), magnetoencephalography (MEG), and positron emission tomography (PET), etc. With high-performance computational tools and various machine learning (ML) and deep learning (DL) methods, these modalities have discovered exciting possibilities for identifying and diagnosing neurological disorders. This study follows a computer-aided diagnosis methodology, leading to an overview of pre-processing and feature extraction techniques. The performance of existing ML and DL approaches for detecting NDs is critically reviewed and compared in this article. A comprehensive portion of this study also shows various modalities and disease-specified datasets that detect and records images, signals, and speeches, etc. Limited related works are also summarized on NDs, as this domain has significantly fewer works focused on disease and detection criteria. Some of the standard evaluation metrics are also presented in this study for better result analysis and comparison. This research has also been outlined in a consistent workflow. At the conclusion, a mandatory discussion section has been included to elaborate on open research challenges and directions for future work in this emerging field.
International Journal On February-2022
Unknown Object Detection Using a One-Class Support Vector Machine for a Cloud–Robot System
Sensors
SCIE | Q1 | Impact Factor: 3.576
Abstract
Inter-robot communication and high computational power are challenging issues for deploying indoor mobile robot applications with sensor data processing. Thus, this paper presents an efficient cloud-based multirobot framework with inter-robot communication and high computational power to deploy autonomous mobile robots for indoor applications. Deployment of usable indoor service robots requires uninterrupted movement and enhanced robot vision with a robust classification of objects and obstacles using vision sensor data in the indoor environment. However, state-of-the-art methods face degraded indoor object and obstacle recognition for multiobject vision frames and unknown objects in complex and dynamic environments. From these points of view, this paper proposes a new object segmentation model to separate objects from a multiobject robotic view-frame. In addition, we present a support vector data description (SVDD)-based one-class support vector machine for detecting unknown objects in an outlier detection fashion for the classification model. A cloud-based convolutional neural network (CNN) model with a SoftMax classifier is used for training and identification of objects in the environment, and an incremental learning method is introduced for adding unknown objects to the robot knowledge. A cloud–robot architecture is implemented using a Node-RED environment to validate the proposed model. A benchmarked object image dataset from an open resource repository and images captured from the lab environment were used to train the models. The proposed model showed good object detection and identification results. The performance of the model was compared with three state-of-the-art models and was found to outperform them. Moreover, the usability of the proposed system was enhanced by the unknown object detection, incremental learning, and cloud-based framework.
International Journal On December-2021
A Comprehensive Survey on Deep-Learning-Based Breast Cancer Diagnosis
cancers
SCIE | Q1 | Impact Factor: 6.639
Abstract
Breast cancer is now the most frequently diagnosed cancer in women, and its percentage is gradually increasing. Optimistically, there is a good chance of recovery from breast cancer if identified and treated at an early stage. Therefore, several researchers have established deep-learning-based automated methods for their efficiency and accuracy in predicting the growth of cancer cells utilizing medical imaging modalities. As of yet, few review studies on breast cancer diagnosis are available that summarize some existing studies. However, these studies were unable to address emerging architectures and modalities in breast cancer diagnosis. This review focuses on the evolving architectures of deep learning for breast cancer detection. In what follows, this survey presents existing deep-learning-based architectures, analyzes the strengths and limitations of the existing studies, examines the used datasets, and reviews image pre-processing techniques. Furthermore, a concrete review of diverse imaging modalities, performance metrics and results, challenges, and research directions for future researchers is presented.
International Journal On November-2021
U-Vectors: Generating Clusterable Speaker Embedding from Unlabeled Data
Applied Sciences
SCIE | Q2 | Impact Factor: 2.679
Abstract
Speaker recognition deals with recognizing speakers by their speech. Most speaker recognition systems are built upon two stages, the first stage extracts low dimensional correlation embeddings from speech, and the second performs the classification task. The robustness of a speaker recognition system mainly depends on the extraction process of speech embeddings, which are primarily pre-trained on a large-scale dataset. As the embedding systems are pre-trained, the performance of speaker recognition models greatly depends on domain adaptation policy, which may reduce if trained using inadequate data. This paper introduces a speaker recognition strategy dealing with unlabeled data, which generates clusterable embedding vectors from small fixed-size speech frames. The unsupervised training strategy involves an assumption that a small speech segment should include a single speaker. Depending on such a belief, a pairwise constraint is constructed with noise augmentation policies, used to train AutoEmbedder architecture that generates speaker embeddings. Without relying on domain adaption policy, the process unsupervisely produces clusterable speaker embeddings, termed unsupervised vectors (u-vectors). The evaluation is concluded in two popular speaker recognition datasets for English language, TIMIT, and LibriSpeech. Also, a Bengali dataset is included to illustrate the diversity of the domain shifts for speaker recognition systems. Finally, we conclude that the proposed approach achieves satisfactory performance using pairwise architectures.
International Journal On September-2021
User Authentication Based on Handwriting Analysis of Pen-Tablet Sensor Data Using Optimal Feature Selection Model
Future Internet 2021, 13(9), 231
ESCI | Q2
Abstract
Handwriting analysis is playing an important role in user authentication or online writer identification for more than a decade. It has a significant role in different applications such as e-security, signature biometrics, e-health, gesture analysis, diagnosis system of Parkinson’s disease, Attention-deficit/hyperactivity disorders, analysis of vulnerable people (stressed, elderly, or drugged), prediction of gender, handedness and so on. Classical authentication systems are image-based, text-dependent, and password or fingerprint-based where the former one has the risk of information leakage. Alternatively, image processing and pattern-analysis-based systems are vulnerable to camera attributes, camera frames, light effect, and the quality of the image or pattern. Thus, in this paper, we concentrate on real-time and context-free handwriting data analysis for robust user authentication systems using digital pen-tablet sensor data. Most of the state-of-the-art authentication models show suboptimal performance for improper features. This research proposed a robust and efficient user identification system using an optimal feature selection technique based on the features from the sensor’s signal of pen and tablet devices. The proposed system includes more genuine and accurate numerical data which are used for features extraction model based on both the kinematic and statistical features of individual handwritings. Sensor data of digital pen-tablet devices generate high dimensional feature vectors for user identification. However, all the features do not play equal contribution to identify a user. Hence, to find out the optimal features, we utilized a hybrid feature selection model. Extracted features are then fed to the popular machine learning (ML) algorithms to generate a nonlinear classifier through training and testing phases. The experimental result analysis shows that the proposed model achieves more accurate and satisfactory results which ensure the practicality of our system for user identification with low computational cost.
International Journal On August-2021
Brain-Computer Interface: Advancement and Challenges (Review Paper)
Sensors 2021, 21(17), 5746
SCIE | Q1 | Impact Factor: 3.576
Abstract
Brain-Computer Interface (BCI) is an advanced and multidisciplinary active research domain based on neuroscience, signal processing, biomedical sensors, hardware, etc. Since the last decades, several groundbreaking research has been conducted in this domain. Still, no comprehensive review that covers the BCI domain completely has been conducted yet. Hence, a comprehensive overview of the BCI domain is presented in this study. This study covers several applications of BCI and upholds the significance of this domain. Then, each element of BCI systems, including techniques, datasets, feature extraction methods, evaluation measurement matrices, existing BCI algorithms, and classifiers, are explained concisely. In addition, a brief overview of the technologies or hardware, mostly sensors used in BCI, is appended. Finally, the paper investigates several unsolved challenges of the BCI and explains them with possible solutions
International Conference On July-2021
A Cloud-based Robot Framework for Indoor Object Identification Using Unsupervised Segmentation Technique and Convolution Neural Network (CNN)
The 34th International Conference on Industrial, Engineering & Other Applications of Applied Intelligent Systems (IEA/AIE 2021), Springer
SCOPUS
Abstract
Nowadays autonomous indoor mobile robot is getting more attention in many application areas. A cloud-based multi-robot framework provides highspeed data processing and inter robot’s communication efficiently for the indoor mobile robot system. However, efficient detection and recognition of the environmental objects are vital issues for indoor mobile robots. Thus, this paper proposes a cloud-based multi-robot system, where Indoor objects are detected using a new unsupervised object segmentation model and object identification using cloud-based Convolutional Neural Networks (CNN) model. In the object segmentation model, a segmentation algorithm is developed with the combination of Canny edge detection, Floodfill, and BoundingBox image processing technique for efficiently segmenting the objects of the indoor environment. After detecting objects, a cloud-based CNN model with SoftMax classifier is used for classifying objects. Besides, an iterative learning is introduced in our proposed model for identifying unknown objects. Some indoor images captured by the camera are used to test the proposed system. To validate the proposed model, a benchmarked object image dataset from an open resource repository is used in this paper to train the CNN model. The model shows good object detection and identification result and the cloud-based framework enhance the usability of the proposed system.
International Conference On February-2021
Analysis on Handwriting Using Pen-Tablet for Identification of Person and Handedness
2021 International Conference on Information and Communication Technology for Sustainable Development (ICICT4SD)
Abstract
International Conference On January-2021
User Authentication Through Pen Tablet Data Using Imputation and Flatten Function
3rd IEEE International Conference on Knowledge Innovation and Invention 2020 (IEEE ICKII 2020)
Abstract
--
International Conference On January-2021
Efficient Frequency Domain Feature Extraction Model using EPS and LDA for Human Activity Recognition
3rd IEEE International Conference on Knowledge Innovation and Invention 2020 (IEEE ICKII 2020)
Abstract
--
International Journal On December-2020
A Robust Feature Extraction Model for Human Activity Characterization using 3-axis Accelerometer and Gyroscope Data
Sensors 2020, 20, 6990
SCIE | Q1 | Impact Factor: 3.27
Abstract
International Journal On June-2020
Smoke Object Segmentation and the Dynamic Growth Feature Model for Video-Based Smoke Detection Systems
Symmetry
SCIE | Q2 | Impact Factor: 2.645
Abstract
This article concerns smoke detection in the early stages of a fire. Using the computer-aided system, the efficient and early detection of smoke may stop a massive fire incident. Without considering the multiple moving objects on background and smoke particles analysis (i.e., pattern recognition), smoke detection models show suboptimal performance. To address this, this paper proposes a hybrid smoke segmentation and an efficient symmetrical simulation model of dynamic smoke to extract a smoke growth feature based on temporal frames from a video. In this model, smoke is segmented from the multi-moving object on the complex background using the Gaussian’s Mixture Model (GMM) and HSV (hue-saturation-value) color segmentation to encounter the candidate smoke and non-smoke regions in the preprocessing stage. The preprocessed temporal frames with moving smoke are analyzed by the dynamic smoke growth analysis and spatial-temporal frame energy feature extraction model. In dynamic smoke growth analysis, the temporal frames are segmented in blocks and the smoke growth representations are formulated from corresponding blocks. Finally, the classifier was trained using the extracted features to classify and detect smoke using a Radial Basis Function (RBF) non-linear Gaussian kernel-based binary Support Vector Machine (SVM). For validating the proposed smoke detection model, multi-conditional video clips are used. The experimental results suggest that the proposed model outperforms state-of-the-art algorithms.
International Conference On May-2020
EEG Motor Signal Analysis-Based Enhanced Motor Activity Recognition Using Optimal De-noising Algorithm
Proceedings of International Joint Conference on Computational Intelligence. IJCCI 2019. Algorithms for Intelligent Systems. Springer, Singapore
Abstract
Brain–computer interface (BCI) technology provides a communication pathway using the human brain motor imaginary signal to develop applications like robotic hands and automated wheelchair, which are useful for the people who come with several motor disabilities. However, the signals which are acquired in a non-invasive approach come with various types of artifacts which badly effect on the accuracy of the prediction. For the above purposes, this paper proposes a model of human motor activity recognition using electroencephalogram (EEG) signal, in which three major time–frequency domain de-noising algorithms, i.e., Discrete Wavelet Transform (DWT), Empirical Mode Decomposition (EMD) and Savitzky–Golay filter are adopted for de-noising the signals acquired in a non-invasive approach. In this paper, initially, the EEG signals are de-noised using de-noising algorithm. After that, important statistical features from selected EEG channels of the dataset are extracted. Finally, the Support Vector Machine (SVM) classification algorithm is used for classifying particular motor activities. Those three major time–frequency domain de-noising algorithms are compared based on five comparison metrics, i.e., mean squared error (MSE), mean absolute error (MAE), signal-to-noise ratio (SNR), peak signal-to-noise ratio (PSNR), cross-correlation and the classification accuracy using SVM classification. For evaluating the proposed model, an online benchmarked dataset with four classes of motor activities has been used. Among the four classes, the first two classes are used in this research. Those are ‘Left-hand movement’ as Class 1 and ‘Right-hand movement’ as Class 2. The experimental result, the DWT-based de-noising method, shows optimal performance.
International Journal On May-2020
A novel approach of cursive signature generation for personal identity
International Journal of Computer Applications in Technology
SCOPUS
Abstract
Signatures are used in many situations of daily life. In many non-English-speaking countries like Japan, people are generally not familiar with English and they do not have English signatures in most cases. However, many of them want to learn how to write English in cursive and hope to have an English signature. Thus, this study proposes a technique for generating an English signature. We use a cubic Bezier curve for the cursive connection and an affine transformation to modify the input characters. Modifications were made in the slant, scale, space between the characters and line emphasis. In addition, we added some decorative functions, such as putting a circle around the signature. The system also provided an animation to teach users how to write cursive English and the order of strokes used in a signature. As a result, we were able to provide system-generated signatures that satisfied the users.
International Conference On February-2020
Hand movement activity based approach for virtual keyboard input
Abstract
Gesture-based input devices are contributing to advanced technology for the protection of confidential information and the convenience of day-to-day operations. There are many touches and non-touch devices that can handle user interface elements using a natural hand gesture such as a tablet display, smartphone, tablet PC with a sensor. However, hand gesture processing is challenging and requires overhead image processing due to tracking, segmentation, and recognition. Therefore, we propose a virtual keyboard input system based on the hand gesture movement that allows the user to input the character with ease and intuition. To do this, a wearable device like the Myo armband is used to achieve hand movement and detect activity by analyzing accelerometer, gyroscope, and electromyography (EMG) signals. A wavelet de-noising the technique is used to remove the noise from input signals and to explore the potential features using the envelope spectrum for the accelerometer and gyroscope, and cepstrum analysis for the EMG signals. Finally, the support vector machine is used to train and detect the gesture to perform character input. In order to validate the proposed model, signal information is obtained from predefined gestures i.e., “double-tapâ€, “hold fistâ€, “wave leftâ€, “wave right†and “spread finger†of different respondents for different input actions such as “input a characterâ€, “change characterâ€, “delete a characterâ€, “line breakâ€, “space characterâ€. The experimental results show the superiority of hand gesture recognition and accuracy of character input compared to state-of-the-art systems.
International Journal On January-2020
Hand Gesture Recognition-Based Non-Touch Character Writing System on a Virtual Keyboard
Multimedia Tools and Applications
SCIE | Q1 | Impact Factor: 2.1
Abstract
The non-touch system is a modern approach of computer-interface technology capable of revolutionizing human-computer interaction. The interface allows the user to input data and interact with a human, machine or robot in an uncontrolled environment, treatment or industrial life. However, it is challenging to input data into the machine and interact with man and machine with a variety of complexities such as cluttered environment, gesture tracking, and speed. There are many evolving systems, for example, aerial handwriting, sign language recognition, and finger alphabet recognition require substantial effort for all character learning and overhead processing, thence the accuracy of the classification is reduced. Therefore, this paper proposes a non-touch character writing system that allows users to interact and manage the on-screen virtual keyboards in a secure and healthy way by recognizing few hand gestures. We divide this work into two parts: a) hand gesture recognition; and b) gestural flick input using a virtual keyboard. A user-friendly keyboard interface is displayed on the monitor, which uses a flick input method. A deep learning method with CNN is used to extract the features of a gesture. To determine these features, color segmentation is used to detect the hand; color pixels can be obtained by extracting a particular HSV (hue, saturation, value) and applying threshold masking to the input image. Finally, a support vector machine is used to give a more accurate classification of the hand gestures. The user uses a gestural flick input system to perform non-touch character input and enters the character by viewing the virtual keyboard. The character input is executed based on the recognition of the user’s hand gestures. Character input is evaluated based on the average classification accuracy of hand gestures and character recognition, and the accuracy and speed of input. Then, the system is compared with the state-of-the-art algorithms. The experimental results show that the proposed system can recognize seven typical gesture functions and input characters with 97.93% accuracy, which demonstrate the superiority compared to the state-of-the-art algorithms.
International Journal On January-2020
Enhanced Human Activity Recognition Based on Smartphone Sensor Data Using Hybrid Feature Selection Model
Sensors
SCIE | Q2 | Impact Factor: 3.03
Abstract
Human activity recognition (HAR) techniques are playing a significant role in monitoring the daily activities of human life such as elderly care, investigation activities, healthcare, sports, and smart homes. Smartphones incorporated with varieties of motion sensors like accelerometers and gyroscopes are widely used inertial sensors that can identify different physical conditions of human. In recent research, many works have been done regarding human activity recognition. Sensor data of smartphone produces high dimensional feature vectors for identifying human activities. However, all the vectors are not contributing equally for identification process. Including all feature vectors create a phenomenon known as ‘curse of dimensionality’. This research has proposed a hybrid method feature selection process, which includes a filter and wrapper method. The process uses a sequential floating forward search (SFFS) to extract desired features for better activity recognition. Features are then fed to a multiclass support vector machine (SVM) to create nonlinear classifiers by adopting the kernel trick for training and testing purpose. We validated our model with a benchmark dataset. Our proposed system works efficiently with limited hardware resource and provides satisfactory activity identification.
International Conference On January-2020
Demographic Factors of Cybersecurity Awareness in Bangladesh
2019 5th International Conference on Advances in Electrical Engineering (ICAEE), Dhaka, Bangladesh, 26-28 Sept. 2019
Abstract
The growing rate of cybercrime has turned into a crucial concern that makes people more afraid of cyberwar than a nuclear weapon and climate change. An alarming awareness level has been observed in this study which is not satisfactory. A significant percentage of the population is unaware of cybersecurity policies and practices. Even the relevant organizations, as well as the government, are not concerned about the issues of cybercrime. A significant variation is essential to enrich the cybersecurity policies and practices. Furthermore, it needs to be monitored and restructured with the pace of time. This pilot study is designed to consider a comprehensive investigation regarding the level of cybercrime awareness amongst Bangladeshi people. The survey has been carried out through both the online and offline responses against a wellprepared questionnaire. Additionally, T-test, mean score and ANOVA tests were accompanied for detailed analysis. Based on this study, it is recommended that the development and implementation of a massive cybersecurity awareness program are urgently required that would immune the citizens from cyber threats.
International Conference On January-2020
Discriminant Feature Extraction using Disease Segmentation for automatic Leaf Disease Diagnosis
International Conference on Computing Advancements (ICCA 2020).
ACM
Abstract
Plant leaf disease is a serious issue in agriculture worldwide. Image processing-based leaf disease detection and diagnosis approaches are becoming popular. The distinguishable features of leaf image are important to improve the performance of disease detection and diagnosis model. However, proper segmentation of the diseased part of a leaf helps to extract discriminant features. From this point of view, this paper presents an efficient leaf disease diagnosis model, where discriminant features are extracted from disease segmented leaf images using optimal segmentation algorithms. In this paper, the disease and healthy parts of the leaf are separated using segmentation techniques. After that, different distinguishable features are extracted based on color and intensity value of image pixels of leaf image. Finally, the Multiclass Support vector machine (MC-SVM) with RBF Gaussian kernel is used for detecting and diagnosis the leaf disease. To validate the proposed model, an online benchmarked leaf images dataset named "PlantVillage-Dataset" is used. The performance evaluation of the proposed model shows a satisfactory result of disease detection performance. Moreover, three well-known image segmentation algorithms are used and an optimal segmentation method is investigated to find better accuracy of disease diagnosis. According to the experimental result, the optimal results are observed for the leaf disease diagnosis using the Otsu thresholding-based disease segmentation.
International Conference On December-2019
A Novel Dynamic Hand Gesture and Movement Trajectory Recognition model for Non-Touch HRI Interface
2019 IEEE Eurasia Conference on IOT, Communication and Engineering (IEEE ECICE 2019), 3-6 Oct. 2019 Taiwan
Abstract
Efficient Human Robot Interaction (HRI) interface is very much demandable for controlling the semi-autonomous robots. Hand gesture recognition is an effective form of non-touch instruction. Thus, human hand gesture recognition is mostly used technique for HRI. However, in most research, some sensor devices or marker are incorporate with the hand or a large number of hand image and hand gesture sequence is stored and process for gesture recognition in machine learning techniques, which are costly and demand complex computation. From this point of view, an efficient dynamic hand gesture and movement trajectory recognition system is proposed in this paper, which can be used in real-time fashion for effective HRI interface. In the proposed dynamic gesture recognition system, hand images and skeleton information are extracted for Kinect sensor. Hands are segmented from the video frame using a skin color segmentation model from the region of interest (ROI) around the palm position of both hands. The hand open and close states are identified by calculating the position of palm and extreme position of Figure for activating the instruction recognition. The trajectory of segmented hands and the hands open state are considered for formulating the model of gesture with respect to the selected index points of body skeleton. Finally, several gesture models are derived to recognize the instruction during temporal gesture movement. For validating the proposed model, an experimental environment is setup in experimental lab. Ten volunteers are considered and tested the proposed system for six gesture instructions. According to the experiment, the proposed system shows 94.5% average recognition accuracy for dynamic motion instruction identification.
International Conference On December-2019
Smartphone Sensor Based Physical Activity Identification by Using Hardware-Efficient Support Vector Machines for Multiclass Classification
2019 IEEE Eurasia Conference on IOT, Communication and Engineering (IEEE ECICE 2019), 3-6 Oct. 2019 Taiwan
Abstract
Smartphone sensor-based activity identification has been recently received significant attention in versatile applications such as elderly people physical condition monitoring, general health monitoring, disease likelihood and other vital contexts to make human life more productive, secure and sound. Due to sensor derived data popularity, along with smartphone researchers are using other ad-hoc wearable devices like smartwatch, fitness tracker, fitbit for activity data collection. This paper emphasizes on heterogeneous optimal feature selection process based on Sequential Floating Forward Search (SFFS) approach. At the first stage, prominent discriminant features are elected from both time and frequency domain signal in order to create a robust model with better accuracy and generalization capability. Then the prime features are trained by Multiclass Support Vector Machines (SVMs) to identify twelve human activities by analyzing accelerometer-gyroscope sensory data taken from a published dataset. In this paper, SVM is applied to create nonlinear classifiers by adopting the kernel trick. Lastly, we have validated our model with online based benchmark dataset. Our proposed system works efficiently with limited hardware resource and provides satisfactory activity identification with trivial time duration as opposed to neural network.
International Journal On October-2019
Gestural Flick Input-Based Nontouch Interface for Character Input
The Visual Computer (TVCJ)
SCIE | Q2 | Impact Factor: 1.415
Abstract
A non-touch character input is a modern system for communication between humans and computers that can help the user to interact with a computer, a machine, or a robot in unavoidable circumstances or industrial life. There have been many studies in the field of touch and non-touch character input systems (i.e., hand gesture languages), such as aerial handwriting, sign languages, and the finger alphabet. However, many previously developed systems require substantial effort in terms of learning and overhead processing for character recognition. To address this issue, this paper proposes a gesture flick input system that offers a quick and easy input method using a hygienic and safe non-touch character input system. In the proposed model, the position and state of the hands (i.e., open or closed) are recognized to enable flick input and to relocate and resize the on-screen virtual keyboard for the user. In addition, this system recognizes hand gestures that perform certain motion functions, such as delete, add a space, insert a new line, and select language, an approach which reduces the need for recognition of a large number of overhead gestures for the characters. To reduce the image-processing overhead and eliminate the surrounding noise and light effects, body index skeleton information from the Kinect sensor is used. The proposed system is evaluated based on the following factors: (a) character selection, recognition and speed of character input (in Japanese hiragana, English, and numerals); and (b) accuracy of gestures for the motion functions. The system is then compared to state-of-the-art algorithms. A questionnaire survey was also conducted to measure the user acceptance and usability of this system. The experimental results show that the average recognition rates for characters and motion functions were 98.61% and 97.5%, respectively, thus demonstrating the superiority of the proposed model compared to the state-of-the-art algorithms.
International Journal On September-2019
Non-Touch Sign Word Recognition Based on Dynamic Hand Gesture Using Hybrid Segmentation and CNN Feature Fusion
Applied Science, 2019, 9(18), 3790
SCIE | Q1 | Impact Factor: 2.217
Abstract
Hand gesture-based sign language recognition is a prosperous application of human– computer interaction (HCI), where the deaf community, hard of hearing, and deaf family members communicate with the help of a computer device. To help the deaf community, this paper presents a non-touch sign word recognition system that translates the gesture of a sign word into text. However, the uncontrolled environment, perspective light diversity, and partial occlusion may greatly affect the reliability of hand gesture recognition. From this point of view, a hybrid segmentation technique including YCbCr and SkinMask segmentation is developed to identify the hand and extract the feature using the feature fusion of the convolutional neural network (CNN). YCbCr performs image conversion, binarization, erosion, and eventually filling the hole to obtain the segmented images. SkinMask images are obtained by matching the color of the hand. Finally, a multiclass SVM classifier is used to classify the hand gestures of a sign word. As a result, the sign of twenty common words is evaluated in real time, and the test results confirm that this system can not only obtain better-segmented images but also has a higher recognition rate than the conventional ones
International Conference On August-2019
Genetic Algorithm Based Optimal Feature Selection Extracted by Time-Frequency Analysis for Enhanced Sleep Disorder Diagnosis Using EEG Signal
Intelligent Systems Conference (IntelliSys), Advances in Intelligent Systems and Computing, vol. 1038, pp. 881-894 2019, London
SCOPUS
Abstract
Sleep disorders have a significant effect on psychological depression and many other human diseases. Nowadays, technology, as well as innovation, has become an essential and analytical part of the world. Detection of sleep disorders by brain waves has become a dynamic study. From this point of view, this paper proposed a model for detecting sleep disorders using time-frequency analysis based feature extraction model based on EEG (Electroencephalogram) signal. In this proposed method, Empirical mode decomposition (EMD) and wavelet packet transform (WPT) time-frequency analysis techniques are used for extracting effective features, because those techniques are very effective for analyzing the non-stationary signal like EEG. In this research, the EMD and WPT are used to decompose the input signal. In EMD decomposition, up to 9th IMFs (Intrinsic Mode Functions) are decomposed. In WPT decomposition, the EEG signal is decomposed up to third level wavelet coefficient. After the decomposition process, different statistical features are extracted, i.e., Shannon entropy, energy, standard deviation, skewness, and kurtosis. However, identifying the optimal sub-band of time-frequency analysis is very challenging. Thus, the genetic algorithm (GA) is used to select the effective subset of the feature. In the detection process, SVM classifier is used and sleep disorders are classified based on trained knowledge. As a result, the performance of the proposed method is evaluated for various statistical features and to find the optimal features for detecting sleep disorder. According to the experimental results, the proposed model shows improved performance by 4.88% improved classification accuracy.
International Journal On June-2019
Detecting and Learning Unknown Fault States by Automatically Finding the Optimal Number of Clusters for Online Bearing Fault Diagnosis
Applied Science 2019, 9(11), 2326
SCIE | Q1 | Impact Factor: 2.217
Abstract
This paper proposes an online fault diagnosis system for bearings that detect emerging fault modes and then updates the diagnostic system knowledge (DSK) to incorporate information about the newly detected fault modes. New fault modes are detected using k-means clustering along with a new cluster evaluation method, i.e., multivariate probability density function’s cluster distribution factor (MPDFCDF). In this proposed model, a heterogeneous pool of features is constructed from the signal. A hybrid feature selection model is adopted for selecting optimal feature for learning the model with existing fault mode. The proposed online fault diagnosis system detects new fault modes from unknown signals using k-means clustering with the help of proposed MPDFCDF cluster evaluation method. The DSK is updated whenever new fault modes are detected and updated DSK is used to classify faults using the k-nearest neighbor (k-NN) classifier. The proposed model is evaluated using acoustic emission signals acquired from low-speed rolling element bearings with different fault modes and severities under different rotational speeds. Experimental results present that the MPDFCDF cluster evaluation method can detect the optimal number of fault clusters, and the proposed online diagnosis model can detect newly emerged faults and update the DSK effectively, which improves the diagnosis performance in terms of the average classification performance
International Conference On June-2019
Dynamic Hand Gesture Based Sign Word Recognition Using Convolutional Neural Network with Feature Fusion
IEEE INTERNATIONAL CONFERENCE ON KNOWLEDGE INNOVATION AND INVENTION 2019 (IEEE ICKII 2019), Seoul, South Korea
Abstract
International Journal On June-2019
Arm Movement Activity Based User Authentication in P2P Systems
Peer-to-Peer Networking and Applications (Accepted)
SCIE | Q2 | Impact Factor: 2.239
Abstract
User authentication has become an essential security element that enables a wide range of applications in P2P systems for higher security and safety requirements. In previous, many researchers worked on user authentication based on certificates, passwords, and feature-based authentication (e.g. face recognition, fingerprint detection, iris recognition, voice recognition). However, authentication using those technologies may fail because this information can be easily shared among users or synthesized. Also, there are several cyber and cryptography attacks. With the progress of the latest sensor technology, wearable as Microsoft Bands, Fitbit, and Garmin has provided for more information collecting opportunities. From those above point of views, this paper presents a novel user identification system based on the bio signal analysis of arm movement (3-axis accelerometer & 3-axis gyroscope) and electromyography (EMG) signal using Myo armband as a wearable user authentication system in P2P system that identifies users based on the bio-signal of movement of a person's arm. In this study, the gesture and EMG signals are obtained from the sensor and denoised using wavelet denoising algorithm. The denoised signals are analyzed using the envelope and cepstrum analysis for extracting the potential feature vector. Finally, the feature vector is used to train and identify a user using multi-class support vector machine (MC-SVM) with different kernel function for user authentication. For validating the proposed authentication model, signals are obtained from the arm movements, i.e., directions and hand gesture data using acceleration, gyroscope and EMG sensors of several subjects. According to the experimental results, the proposed model shows satisfactory performance. To evaluate the efficiency of the proposed systems, we measure and compare its classification accuracy with state-of-the-art algorithms. And the proposed algorithm outperforms with others.
International Conference On May-2019
A lossless speech signal compression technique
1st International Conference on Advances in Science, Engineering and Robotics Technology(ICASERT-2019) (Accepted)
Abstract
International Conference On December-2018
Histogram alternation based digital image compression using base-2 coding
2018 Digital Image Computing: Techniques and Applications (DICTA), Canberra, Australia,
Abstract
The goal of data compression algorithms is to promote the storage and delivery of big images with excellent compression ratio and least distortion. As the number of internet user is growing day by day, the transfer of data is being another significant concern. The storage and the use of an uncompressed picture are very costly and time-consuming. There are many techniques such as Arithmetic coding, Run-length coding, Huffman coding, Shannon-Fano coding used to compress an image. Compression of a picture using the state-of-the-art techniques has a high impact. However, The compression ratio and transfer speed do not satisfy the current demand. This article proposes a new histogram alternation based lossy image compression using Base-2 coding. It increases the probabilities of an image by doing a little bit of change to its pixels level which helps to reduce code-word. This algorithm uses less storage space and works at high-speed to encode and decode an image. Average code length, compression ratio, mean square error and pick signal to noise ratio are used to estimate this method. The proposed method demonstrates better performance than the state-of-theart techniques.
International Conference On December-2018
PCA and ICA Based Hybrid Dimension Reduction Model for Cardiac Arrhythmia Disease Diagnosis
IEEE 10th International Conference on Humanoid, Nanotechnology, Information Technology, Communication and Control, Environment, and Management, Baguio, Philippines
Abstract
Abstract—An arrhythmia is a fluctuation in the continuous beat of the heart (i.e., anomalous rhythm). Arrhythmia is considered a hazardous disease causing genuine medical problems in patients, when left untreated. For saving lives, early diagnosis of arrhythmias would be very conducive. The P-QRS-T wave of the Electrocardiogram (ECG) signal illustrates the cardiac function. However, it is a tough task to extract the discriminant information from a large number of data of ECG signal. In this perspective, this study exhibits a novel approach for diagnosing diseases related to cardiac arrhythmia. In this proposed model, a hybrid dimension reduction model including Independent and Principal Component Analysis (ICA, PCA) are introduced and machine learning features are extracted for disease diagnosis. The original ECG data are splitted into several windows and consider as input of dimension reduction process. After completing the ICA and PCA process, the different components of ICA and PCA are used for feature extraction. Finally, the Multi-Class Support Vector Machine (MCSVM) is used for training and identifying the disease. For evaluating the proposed method, MIT-BIH dataset is used. According to the experiment, the proposed model shows better classification accuracy using the first components of ICA and PCA algorithms, which is 98.67%.
International Conference On November-2018
Optimal IMF Selection of EMD for Sleep Disorder Diagnosis using EEG Signals
In The 3rd International Conference on Applications in Information Technology (ICAIT’18)
Abstract
Sleep disorders has a vital effect on mental depression and many other diseases of human body. Diagnosing the sleep disorder in an early curable stage may help to provide better treatment and save the life. The EEG (Electroencephalogram) signal is one of the most uses bio signal for capturing brain activities to detect and diagnosis the sleep disorders. Empirical mode decomposition (EMD) is an efficient time-frequency data analysis technique for diagnosing disease by analyzing EEG signal. However, it is a challenging issue to select the optimal intrinsic mode functions (IMFs) of Empirical mode decomposition (EMD) for extracting discriminant properties of EEG signals to diagnosis the sleep disorder. From this point of view, this paper presents a model to select optimal IMF of EMD for diagnosing the sleep disorder using EEG brain signal. In this proposed model, EMD is applied to decompose and analyze EEG signal for extracting biomarker/feature of sleep disorders. During the EMD decomposition process, different levels of IMF are extracted and features, i.e., Shannon Entropy, Spectral Entropy, Standard deviation, Skewness and Kurtosis are calculated from those IMFs for detecting the sleep disorders. In identification process, the multiclass support vector machine (MC-SVM) classification algorithm is used and sleep disorders are classified based on trained knowledge. Finally, the performance of proposed model is evaluated for different IMFs of EMD and find the optimal IMF for sleep disorder diagnosis. For evaluating the proposed model, a benchmark dataset including 4 types of data such as Apnea, REM, PLM and healthy subjects are used in experiment. According to the experimental result, the proposed model achieves the optimal classification performance for IMF 8, i.e., 93.24% average classification accuracy.
International Conference On November-2018
Human Machine Interaction based on Hand Gesture Recognition using Skeleton Information of Kinect Sensor
In The 3rd International Conference on Applications in Information Technology (ICAIT’18) (ACM), Aizu-Wakamatsu, Japan
Abstract
The hand gesture provides a natural and intuitive communication medium for the human and machine interaction. Because, it can use in virtual reality, language detection, computer games, and other human-computer or human-machine instruction applications. Currently, the sensor and camera-based application is a field of interest for many researchers. This paper proposes a new hand gesture recognition system using the Kinect sensor's skeleton data, which works in an environment where people do not touch devices or communicate verbally. The proposed model focuses on mainly two modules, namely, hand area and fingertip detection, and hand gesture recognition. The hand area and fingertip are detected by positioning the palm point and find extreme of contour. And, the hand gesture is recognized by measuring the distance between different body indexes of skeleton information. Here, six gestures instructions are considered such as move right to left, move left to right, move up to down, move down to up, open and closed, and also recognize the numeric number using the fingertip. This system is able to detect the presence of hand area and fingers and to recognize different hand gestures. As a result, the average recognition accuracy of different hand gestures and stretched fingers numbers are 95.91% and 96%, respectively.
International Conference On September-2018
Histogram modification based lossy image compression scheme using Huffman coding
4th International Conference on Electrical Engineering and Information & Communication Technology (iCEEiCT 2018)
Abstract
Medical images produce massive volumes of data, and consequently, suitable image compression procedures require to save expense and time of storage and transportation sequentially. The requirement of transferring or storing satellite images is increasing quickly with the evolution of advanced communications and imaging systems. In the digital world, the size of pictures is a major problem when dealing with the storage and transportation necessities. Compression is one of the most primary procedures to address this difficulty. The intention of data compression is to promote the storage and delivery of big images with excellent compression ratio and least distortion. Moreover, the number of internet user is growing day by day speedily. So, transferring of data is being an another significant concern. This article introduces histogram modification based lossy image compression using Huffman coding. A little bit change of pixel’s value is done which is why the number of probabilities of an original image is decreased and the value of probabilities are increased. As a result, Huffman coding uses very few bits in case of encoding and decoding than that of previous. This process provides higher compression ratio and less average code length keeping the same quality of the corresponding original image. Keywords: Lossy compression, Lossless compression, Huffman Coding, PSNR, Compression ratio, SVD coding.
International Conference On September-2018
Hand Gesture Feature Extraction using Deep Convolutional Neural Network for Recognizing American Sign Language
2018 4th International Conference on Frontiers of Signal Processing (ICFSP 2018) France
Abstract
International Conference On June-2018
A Novel Lossless Coding Technique for Image Compression
2018 Joint 7th International Conference on Informatics, Electronics & Vision (ICIEV) and 2018 2nd International Conference on Imaging, Vision & Pattern Recognition (icIVPR)
Abstract
Advanced imaging technology produces a vast amount of data, particularly from the computed tomography modality. Store the imaging data and/or transfer the data through internet is a significant challenge in terms of cost and time. From this point of view, it is obligatory to compress the imaging data effectively. There are many state of the art coding techniques used to compress data such as Huffman coding. This paper proposes a new coding technique for loss-less imaging data compression. In this proposed method, the probability of image data has been assigned by binary sequence with fewer number of bits which provides an improved result than the state of art techniques in terms of compression ratio and average code length.
International Journal On January-2018
Texture Analysis Based Feature Extraction Using Gabor Filter and SVD for Reliable Fault diagnosis of an Induction Motor
International Journal of Information Technology and Management
SCOPUS
Abstract
This paper presents a texture analysis based feature extraction method using a Gabor filter and singular value decomposition (SVD) for reliable fault diagnosis of an induction motor. This method first converts one-dimensional (1D) vibration signal to a two-dimensional (2D) grey-level texture image for each fault signal. Then, the 2D Gabor filter with optimal frequency and orientation values is used to extract a filtered image with distinctive texture information, and SVD is utilised to decompose the Gabor filtered image and select finer singular values of SVD as discriminative features for multi-fault diagnosis. Finally, one-against-all multiclass support vector machines (OAA-MCSVMs) are used as classifiers. In this study, multiple induction motor faults with different noisy conditions are used to validate the proposed fault diagnosis methodology. The experimental results indicate that the proposed method achieves an average classification accuracy of 99.86% and outperforms conventional fault diagnosis algorithms in the fault classification accuracy.
International Journal On December-2017
A video-based smoke detection using smoke flow pattern and spatial-temporal energy analyses for alarm systems
Information Sciences, Volumes 418-419, Pages 91-101, ISSN 0020-0255.
SCI | Impact Factor: 4.832
Abstract
Detecting smoke during the initial stages is vital for preventing fire events. This study proposes a video-based approach for alarm systems that detects smoke based on temporal features extracted from optical smoke flow pattern analysis and spatial-temporal energy analysis. To do this, it considers various optical characteristics such as the diffusion, color, and semi-transparency of smoke. In the proposed model, smoke-colored pixels are identified via masking in the HSV color space and a temporal frame difference is applied. To extract the temporal feature vectors, we propose a new method that determines the optical flow of smoke by using distinguished texture information by applying a Gabor filter bank with preferred orientations. In addition, when applied to an image that has been temporal-differenced, the energy of the spatial frequencies is fed as another feature into the feature vector. Finally, these features are fed to a support vector machine (SVM) to discriminate our data more thoroughly and provide accurate detection of smoke. Experiments are carried out with benchmark datasets, which show that the proposed approach can work effectively without false alarms.
International Conference On December-2017
Reduction of Gesture Feature Dimension for Improving the Hand Gesture Recognition Performance of Numerical Sign Language
20th International Conference On Computer and Information Technology (ICCIT 2017)
Abstract
A major form of non-touch human-computer interaction (HCI) is hand gesture recognition. This is one of the appealing ways to interact with computers and a natural part of how we communicate. However, as a part of HCI, human hand gesture recognition is a challenging issue. From this point of view, this paper presents an effective hand gesture recognition system with hand feature selection for low cost video acquisition device. In this proposed model, hand features are extracted from video frame using discrete wavelet transformation and singular value decomposition. A genetic algorithm with effective fitness function is used to select optimal hand features by eliminating redundant and irrelevant features for improving the recognition performance. Finally, support vector machine is used to recognize the hand gestures for numerical hand gesture accuracy of American Sign Language. The proposed model is validated using a constructed hand gesture dataset. The proposed model is compared with non-feature selection based models, where the feature selection-embedded model outperforms the traditional hand recognition process.
International Conference On December-2017
Faster Convergence and Reduction of Overfitting in Numerical Hand Sign Recognition using DCNN
5th IEEE Region 10 Humanitarian Technology conference 2017
Abstract
Hand signs and signals are the staple form of expression for the hearing and speech impaired people. Human Computer Interaction technology enable people to interact with computer machine using hand gestures. Common sign languages use separate hand signals to communicate different decimals. Recent developments in Deep Convolutional Neural Networks (DCNN) have opened the door to recognize and classify this visual form of gestures more accurately. In this paper, a layer-wise optimized neural network architecture is proposed where batch normalization contributes to faster convergence of training, and introduction of dropout technique mitigates data overfitting. Batch normalization forces each training batch toward zero mean and unit variance, leading to improved flow of gradients through the model and convergence in shorter time. Dropout forces neurons of neural network to regularize, resulting in reduced overfitting. A constructed numerical hand gesture data set is used for validating the claims based on American Sign Language system. The proposed model is shown to surpass other methods in classifying these numerical hand signs successfully.
International Conference On December-2017
Real-Time Feedback-Centric Nurse Calling System with Archive Monitoring using Raspberry Pi
2017 International Conference on Networking, Systems and Security (NSysS 2017), Dhaka, Bangladesh
Abstract
The relationship between nurse and patient is vital as well as vulnerable on which a significant portion of well-being of patient depends. The recent growth of Internet of Things provides the opportunity to maintain this relationship in a more secure and efficient way. From this point of view, this paper presents a system of real-time nurse calling with focus centered on nurse feedback based on patient condition. The system also includes a robust archive monitoring system for review and trend analysis. In the proposed system, patients are able to call for assistance in time of emergency by pressing a designated button. After this, the device will send a real-time message containing information about the patients bed, room, and floor number to the appropriate nurse station. Nurse will respond as soon as the message is delivered, and optionally call for help by pressing designated buttons in the device. The proposed system reduces the delay in response of nurse. Additionally, the interactions are stored in Raspberry Pi database for future analysis to improve the quality of services. The proposed system uses low-cost hardware like ATMEGA328P microcontroller and ENC28J60 Ethernet controller, and Arduino Uno.
International Conference On September-2017
Reduction of Overfitting in Diabetes Prediction Using Deep Learning Neural Network
7th iCatse International Conference on IT Convergence and Security, 2017 (LNEE)
Abstract
Augmented accuracy in prediction of diabetes will open up new frontiers in health prognostics. Data overfitting is a performance-degrading issue in diabetes prognosis. In this study, a prediction system for the disease of diabetes is pre-sented where the issue of overfitting is minimized by using the dropout method. Deep learning neural network is used where both fully connected layers are fol-lowed by dropout layers. The output performance of the proposed neural network is shown to have outperformed other state-of-art methods and it is recorded as by far the best performance for the Pima Indians Diabetes Data Set.
International Conference On September-2017
Retrieving YouTube Video by Sentiment Analysis on User Comment
2017 IEEE International Conference on Signal and Image Processing Applications (ICSIPA),Kuching, 2017, pp. 474-478
Abstract
YouTube is one of the comprehensive video information source on the web where video is uploading continuously in real time. It is one of the most popular site in social media, where users interact with sharing, commenting and rating (like/views) videos. Generally the quality, relevancy and popularity of the video is maintained based on this rating. Sometimes irrelevant and low quality videos ranked higher in the search result due to the number of views or likes, which seems untenable. To minimize this issue, we present a Natural Language processing (NLP) based sentiment analysis approach on user comments. This analysis helps to find out the most relevant and popular video of YouTube according to the search. The effectiveness of the proposed scheme has been proved by a data driven experiment in terms of accuracy of finding relevant, popular and high quality video.
International Conference On July-2017
Genetic Algorithm Based Discriminant Feature Selection for Improved Fault Diagnosis of Induction Motor
The 19th Int. Conf on Artificial Intelligence ICAI-17 (CSREA Press), Las Vegas, Nevada, USA
Abstract
In this paper, we present an efficient model for reliable fault diagnosis of the induction motor. This is now a growing demand for high classification accuracy in fault diagnosis. However, the system performance is highly dependable on superior feature analysis. But, it’s still crucial and computational complex to select discernment features, thus, a new genetic algorithm (GA) with optimum class separability criteria is utilized to find most discriminate features from a hybrid feature vector. For this approach, wavelet packet decomposition (WPD) is applied on Acoustic Emission (AE) fault signal and hybrid statistical features are extracted from a decomposed wavelet packet coefficient, which has maximum energy. GA and Euclidean distance based novel, optimum class separability (OCS) are used to select the optimal low-dimensional feature set from high dimensional feature set. The efficacy of this proposed model, in terms of classification accuracy, is validated by the knearest neighbor (k-NN) classifier. Experimental results show that the proposed model has a superior classification, yielding an average classification accuracy above 98%.
International Conference On February-2017
Efficient Bearing Fault Diagnosis by Extracting Intrinsic Fault Information using Envelope Power Spectrum
2017 IEEE International Conference on Imaging, Vision &Pattern Recognition , Dhaka
Abstract
Early and efficient fault diagnosis of bearing of industrial motor is a modern demand for reducing unexpected breakdown of industrial process. Extracting the intrinsic fault signature in very early stage is important. In this point of view, this paper proposes a fault diagnosis model of industrial bearing including efficient fault signature extraction technique based on narrow band frequency domain analysis of acoustic emission (AE) signal using envelope power spectrum. To do that, AE signals are collected from defective and non-defective bearings under different rotational speeds from industrial-like experimental environment. Envelope power spectrum is calculated from the AE signal and narrow band root mean square (NBRMS) fault features are extracted from defect frequency ranges of the envelope power spectrum. Finally, the k-nearest neighbor (k-NN) classification algorithm is used for identifying the fault of unknown signal and validating the efficiency of the proposed feature extraction model. The experimental result shows that the proposed model outperforms state-of-art algorithms in terms of classification accuracy.
International Journal On January-2017
A Hybrid feature selection scheme based on local compactness and global separability for improving roller bearing diagnostic performance
Artificial Life and Computational Intelligence, Volume 10142 of the series Lecture Notes in Computer Science (LNCS) pp 180-192.
LNCS
Abstract
This paper proposes a hybrid feature selection scheme for identifying the most discriminant fault signatures using an improved class separability criteria—the local compactness and global separability (LCGS)—of distribution in feature dimension to diagnose bearing faults. The hybrid model consists of filter based selection and wrapper based selection. In the filter phase, a sequential forward floating selection (SFFS) algorithm is employed to yield a series of suboptimal feature subset candidates using LCGS based feature subset evaluation metric. In the wrapper phase, the most discriminant feature subset is then selected from suboptimal feature subsets based on maximum average classification accuracy estimation of support vector machine (SVM) classifier using them. The effectiveness of the proposed hybrid feature selection method is verified with fault diagnosis application for low speed rolling element bearings under various conditions. Experimental results indicate that the proposed method outperforms the state-of-the-art algorithm when selecting the most discriminate fault feature subset, yielding 1.4% to 17.74% diagnostic performance improvement in average classification accuracy.
International Journal On January-2017
Reliable Fault Diagnosis of Bearings Using Distance and Density Similarity on an Enhanced k-NN
Artificial Life and Computational Intelligence,Volume 10142 of the series Lecture Notes in Computer Science (LNCS) pp 193-203
LNCS
Abstract
The k-nearest neighbor (k-NN) method is a simple and highly effective classifier, but the classification accuracy of k-NN is degraded and becomes highly sensitive to the neighborhood size k in multi-classification problems, where the density of data samples varies across different classes. This is mainly due to the method using only a distance-based measure of similarity between different samples. In this paper, we propose a density-weighted distance similarity metric, which considers the relative densities of samples in addition to the distances between samples to improve the classification accuracy of standard k-NN. The performance of the proposed k-NN approach is not affected by the neighborhood size k. Experimental results show that the proposed approach yields better classification accuracy than traditional k-NN for fault diagnosis of rolling element bearings.
Book/Book Chapters On January-2017
A Novel Transfer Learning Approach upon Hindi, Arabic, and Bangla Numerals using Convolutional Neural Networks
International Conference On Computational Vision and Bio Inspired Computing (ICCVBIC 2017), 2017 in Tamil Nadu, India
LNAI
Abstract
Increased accuracy in predictive models for handwritten character recognition will open up new frontiers for optical character recognition. Major drawbacks of predictive machine learning models are headed by the elongated training time taken by some models, and the requirement that training and test data be in the same feature space and consist of the same distribution. In this study, these obstacles are minimized by presenting a model for transferring knowledge from one task to another. This model is presented for the recognition of handwritten numerals in Indic languages. The model utilizes convolutional neural networks with backpropagation for error reduction and dropout for data overfitting. The output performance of the proposed neural network is shown to have closely matched other state-of-the-art methods using only a fraction of time used by the state-of-the-arts.
International Journal On December-2016
Acoustic Emission Sensor Network Based Fault Diagnosis of Induction Motors Using a Gabor Filter and Multiclass Support Vector Machines
Ad Hoc & Sensor Wireless Networks (AHSWN), Vol. 34, pp. 273-287
SCIE | Impact Factor: 1.043
Abstract
Reliable and ef cient fault diagnosis of induction motors is an important issue in industrial environments. This paper proposes a method for reli- able fault diagnosis of induction motors using signal processing of acous- tic emission (AE) data, including Gabor ltering and the use of multiclass support vector machines (MCSVMs), where a ZigBee based wireless sensor network (WSN) model is used for ef ciently transmitting AE sig- nals to a diagnosis server. In the proposed fault diagnosis approach, the induction motor’s different state signals are acquired through proper placement of AE sensors. The AE data are sent to a server through the wireless sensor network and decomposed using discrete wavelet transfor- mation (DWT). An appropriate band is then selected using the maximum energy ratio, and a one-dimensional (1D) Gabor lter with various fre- quencies and orientation angles is applied to reduce abnormalities and extract various statistical parameters for generating features. In addition, principal component analysis (PCA) is applied to the extracted features to select the most dominant feature dimensions. Finally, one-against-one multiclass support vector machines (OAA-MCSVMs) are used to classify multiple fault types of an induction motor, where each SVM individually trains with its own features to increase the fault classi cation accuracy of the induction motor. In experiments, the proposed approach achieved an average classi cation accuracy of 99.80%, outperforming conventional fault diagnosis models.
International Conference On December-2016
Feature Selection Techniques for Increasing Reliability of Fault Diagnosis of Bearings
9th International Conference on Electrical and Computer Engineering (ICECE), 2016 ICECE2016, Dhaka
Abstract
Features selection (FS) techniques have an apparent need in many complex engineering applications especially the bearing fault diagnosis of low-speed industrial motor. The main goal of an FS algorithm is to select the most discriminant features subset from a high-dimension features vector that increases the model performance by reducing the redundant and irrelevant fault features. This paper proposes an efficient fault diagnosis model of bearing by incorporating the optimal feature selection approach for increasing the reliability of fault diagnosis of bearing. Also, this paper investigates the feature selection approaches including sequential forward selection (SFS), sequential floating forward selection (SFFS), and genetic algorithm (GA) for identifying the most discriminant subset. The effectiveness of this discriminant features subset is verified with a low-speed bearing fault diagnosis application for identifying bearing failures. The experimental shows up-to-mark diagnosis performance using GA based optimal feature selection method.
Book/Book Chapters On November-2016
Power of Parallelism in .NET, ISBN-13: 978-3-659-94849-7, ISBN-10: 3659948497
LAP LAMBERT Academic Publishing
Abstract
Parallel programming is being seen as the only cost-effective method to improve performance of computer applications. A well-structured parallel application can achieve better performance in terms of execution speed over the sequential execution on existing and upcoming parallel computer architecture. This book named ? Power of Parallelism in .NET, describes the experimental evaluation of different parallel application performance with thread-safe data structure and parallel constructions in .NET framework. It describes different performance issues of parallel application development. Before describing the experimental evaluation, this book describes some methodologies relevant to parallel programming, such as Parallel Computer Architecture, Memory Architectures, Parallel Programming Models, Decomposition, Threading etc. It describes the different APIs in .NET framework and the way of coding for making an efficient parallel application depends on problem definition. Moreover, this book describes Improper Partitioning, Over-subscription, Improper Workloads and more. The parallel application models and the evolving example of parallel programming is well-illustrated in this book.
International Journal On October-2016
Distance and Density Similarity Based Enhanced k-NN Classifier for Improving Fault Diagnosis Performance of Bearings
Shock and Vibration, vol. 2016, Article ID 3843192, 11 pages
SCIE | Impact Factor: 1.857
Abstract
An enhanced -nearest neighbor (-NN) classification algorithm is presented, which uses a density based similarity measure in addition to a distance based similarity measure to improve the diagnostic performance in bearing fault diagnosis. Due to its use of distance based similarity measure alone, the classification accuracy of traditional -NN deteriorates in case of overlapping samples and outliers and is highly susceptible to the neighborhood size, . This study addresses these limitations by proposing the use of both distance and density based measures of similarity between training and test samples. The proposed -NN classifier is used to enhance the diagnostic performance of a bearing fault diagnosis scheme, which classifies different fault conditions based upon hybrid feature vectors extracted from acoustic emission (AE) signals. Experimental results demonstrate that the proposed scheme, which uses the enhanced -NN classifier, yields better diagnostic performance and is more robust to variations in the neighborhood size, .
Book/Book Chapters On August-2016
Multi-core Accelerated Discriminant Feature Selection for Real-time Bearing Fault Diagnosis
Trends in Applied Knowledge-Based Systems and Data Science, LNCS, vol. 9799, pp 645-656, Morioka, Japan.
Springer - Lecture Notes
Abstract
This paper presents a real-time and reliable bearing fault diagnosis scheme for induction motors with optimal fault feature distribution analysis based discriminant feature selection. The sequential forward selection (SFS) with the proposed feature evaluation function is used to select the discriminative feature vector. Then, the k-nearest neighbor (k-NN) is employed to diagnose unknown fault signals and validate the effectiveness of the proposed feature selection and fault diagnosis model. However, the process of feature vector evaluation for feature selection is computationally expensive. This paper presents a parallel implementation of feature selection with a feature evaluation algorithm on a multi-core architecture to accelerate the algorithm. The optimal organization of processing elements (PE) and the proper distribution of feature data into memory of each PE improve diagnosis performance and reduce computational time to meet real-time fault diagnosis.
International Journal On May-2016
A Hybrid Feature Selection Scheme for Reducing Diagnostic Performance Deterioration Caused by Outliers in Data-Driven Diagnostics
IEEE Transaction on Industrial Electronics (TIE), VOL. 63, NO. 5
SCI | Impact Factor: 7.05
Abstract
In practice, outliers, defined as data points that are distant from the other agglomerated data points in the same class, can seriously degrade diagnostic performance. To reduce diagnostic performance deterioration caused by outliers in data-driven diagnostics, an outlier-insensitive hybrid feature selection (OIHFS) methodology is developed to assess feature subset quality. In addition, a new feature evaluation metric is created as the ratio of the intraclass compactness to the interclass separability estimated by understanding the relationship between data points and outliers. The efficacy of the developed methodology is verified with a fault diagnosis application by identifying defect-free and defective rolling element bearings under various conditions.
International Journal On January-2016
Multi-core Accelerated Discriminant Feature Selection for Real-time Bearing Fault Diagnosis
Trends in Applied Knowledge-Based Systems and Data Science,vol. 9799, pp 645-656.
LNCS
Abstract
This paper presents a real-time and reliable bearing fault diagnosis scheme for induction motors with optimal fault feature distribution analysis based discriminant feature selection. The sequential forward selection (SFS) with the proposed feature evaluation function is used to select the discriminative feature vector. Then, the k-nearest neighbor (k-NN) is employed to diagnose unknown fault signals and validate the effectiveness of the proposed feature selection and fault diagnosis model. However, the process of feature vector evaluation for feature selection is computationally expensive. This paper presents a parallel implementation of feature selection with a feature evaluation algorithm on a multi-core architecture to accelerate the algorithm. The optimal organization of processing elements (PE) and the proper distribution of feature data into memory of each PE improve diagnosis performance and reduce computational time to meet real-time fault diagnosis.
International Journal On January-2016
A Two Dimensional Fault Diagnosis Model of Induction Motors using Gabor Filter on Segmented Images
International Journal of Control and Automation, Vol. 9, No. 1
SCOPUS
Abstract
Image segmentation has received extensive attention due to the use of high-level descriptions of image content. This paper proposes a fault diagnosis model using a Gabor filter on segmented two-dimensional (2D) gray-level images. The proposed approach first converts time domain AE signals into 2D gray-level images to exploit texture information from the converted images. 2D discrete wavelet transform (DWT) is then applied to select appropriate (vertical) texture information and reconstructed it into an image. The reconstructed image is segmented into a number of sub-images depending on the segment size and a Gabor filter is applied on each sub-image. Finally, feature vectors are extracted from the Gabor-filtered sub-images and utilized as inputs in a one-against-all multiclass support vector (OAA-MCSVM) to identify each fault in an induction motor. In this study, multiple bearing defects under various segment sizes are utilized to validate the effectiveness of the proposed method. Experimental results indicate that the proposed model outperforms conventional Gabor-filter-based 2D fault diagnosis algorithms in classification accuracy, exhibiting a 97 % average classification accuracy for 64×64 segmented images.
International Journal On January-2016
Discriminant feature distribution analysis-based hybrid feature selection for online bearing fault diagnosis in induction motors
Journal of Sensors
SCIE | Impact Factor: 2.057
Abstract
Optimal feature distribution and feature selection are of paramount importance for reliable fault diagnosis in induction motors. This paper proposes a hybrid feature selection model with a novel discriminant feature distribution analysis-based feature evaluation method. The hybrid feature selection employs a genetic algorithm- (GA-) based filter analysis to select optimal features and a -NN average classification accuracy-based wrapper analysis approach that selects the most optimal features. The proposed feature selection model is applied through an offline process, where a high-dimensional hybrid feature vector is extracted from acquired acoustic emission (AE) signals, which represents a discriminative fault signature. The feature selection determines the optimal features for different types and sizes of single and combined bearing faults under different speed conditions. The effectiveness of the proposed feature selection scheme is verified through an online process that diagnoses faults in an unknown AE fault signal by extracting only the selected features and using the -NN classification algorithm to classify the fault condition manifested in the unknown signal. The classification performance of the proposed approach is compared with those of existing state-of-the-art average distance-based approaches. Our experimental results indicate that the proposed approach outperforms the existing methods with regard to classification accuracy.
International Conference On December-2015
Many-core Accelerated Local Outlier Factor (LOF) based Classifier in Bearing Fault Diagnosis
18th International Conference on Computer and Information Technology (ICCIT)
Abstract
This paper proposes a feature extraction, selection, and classification based bearing fault diagnosis methodologies using acoustic emission (AE) signal. First, a set of statistical, time-domain, and frequency domain features are extraction from AE signal. Of these features, some features having more informative data to distinguish faults are selected. Finally, a classifier based on local outlier factor (LOF) is used to detect faults of bearing. LOF consists of calculating distances of each input will all the training features, thus having a lot of computations. To reduce execution time, this paper implemented the LOF on a data parallel many-core architecture. Experimental results showed that, many-core implementation of LOF algorithm is more than 923× faster than sequential implementation.
International Journal On March-2015
A Centroid-GPS Model to Improving Positioning Accuracy for a Sensitive Location-Based System
Lecture Notes in Electrical Engineering Volume 331
SCOPUS
Abstract
This paper proposes a centroid global positioning system (GPS) model to improve the positioning accuracy of low-cost GPS receivers of a sensitive location-based system. The proposed model estimates the precise movement position by a centroid sum of the individual improved positions of three GPS receivers. Each GPS receiver’s position is improved by using a direction and velocity averaging technique based on combining the vehicle movement direction, velocity averaging, and distance between the waypoints of each GPS receiver using coordinate data (latitude, longitude, time, and velocity). Finally, the precise position is estimated by calculating a triangular centroid sum with distance threshold of the improved positions of three GPS receivers. In order to evaluate the performance of the proposed approach, we used three GARMIN GPS 19x HVS receivers attached to a car and plotted the processed data in Google map. The proposed approach resulted in an improved accuracy of about 2–12 m compared to the original GPS receivers. In addition, we compared the proposed approach to two other state-of-the-art methods. The experimental results show that the proposed approach outperforms the conventional methods in terms of positioning accuracy.
Book/Book Chapters On January-2015
Maximum class separability-based discriminant feature selection using a GA for reliable fault diagnosis of induction motors
Lecture Notes in Artificial Intelligence LNAI
SCOPUS
Abstract
Reliable fault diagnosis in bearing elements of induction motors, with high classification performance, is of paramount importance for ensuring steady manufacturing. The performance of any fault diagnosis system largely depends on the selection of a feature vector that represents the most distinctive fault attributes. This paper proposes a maximum class separability (MCS) feature distribution analysis-based feature selection method using a genetic algorithm (GA). The MCS distribution analysis model analyzes and selects an optimal feature vector, which consists of the most distinguishing features from a high dimensional feature space, for reliable multi-fault diagnosis in bearings. The high dimensional feature space is an ensemble of hybrid statistical features calculated from time domain analysis, frequency domain analysis, and envelope spectrum analysis of the acoustic emission (AE) signal. The proposed maximum class separability-based objective function using the GA is used to select the optimal feature set. Finally, k-nearest neighbor (k-NN) algorithm is used to validate our proposed approach in terms of the classification performance. The experimental results validate the superior performance of our proposed model for different datasets under different motor rotational speeds as compared to conventional models that utilize (1) the original feature vector and (2) a state-of-the-art average distance-based feature selection method.
International Conference On January-2015
Fault Diagnosis of an Induction Motor Using a Gabor Filter and Singular Value Decomposition Based Feature Extraction Method
2015 International Conference on Platform Technology and Service (PlatCon-15)
Abstract
This paper presents an efficient fault diagnosis method of an induction motor using two dimensional (2D) feature extraction including a Gabor filter and singular valued decomposition (SVD). Vibration signals are first converted to gray level images, then a 2D Gabor filter and SVD are applied to extract distinctive texture features. Finally, a multiclass support vector machine (MCSVM) is used as a classifier, where each SVM is individually trained with its own feature vector. Experimental results show that the proposed fault diagnosis model outperforms conventional state-of-the-art fault diagnosis algorithms in terms of classification accuracy, yielding an average classification accuracy of 99.86%.
International Conference On December-2014
Reliable Fault Diagnosis of Induction Motors using 2D Signal Processing Techniques
8th International Conference on Electrical and Computer Engineering (ICECE 2014), Dhaka
Abstract
International Journal On October-2014
Accelerating the formant synthesis of haegeum sounds using a general-purpose graphics processing unit
Multimedia Tools and Applications, Springer
SCIE | Impact Factor: 1054
Abstract
Sound synthesis is recently indispensable with sophisticated audio effects for mimicking rich and natural sounds of the musical instruments, and thus sound synthesis acceleration has been an urgent issue. The formant synthesis is employed to produce the various single notes of the haegeum, a representative traditional Korean bowed string instrument. In this study, the formant synthesis process using multiple pairs of digital resonators and band-pass filters is accelerated with the power of a general-purpose graphics processing unit (GPGPU). This paper compares the performance of the proposed GPGPU-based parallel approach with the CPU-based sequential approach in order to validate the effectiveness of the proposed massively parallel method. Experimental results indicate that the proposed parallel approach achieves at least 79 times speedup over the CPU-based approach by exploiting the massive parallelism inherent in the formant sound synthesis algorithm.
International Conference On July-2014
A New Cooperative-GPS Approach to Improve GPS Positioning Accuracy for Location Based System
The 2nd FTRA International Conference on Ubiquitous Computing Application and Wireless Sensor Network (UCAWSN-14)
Abstract
International Conference On July-2014
Reliable Fault Diagnosis of Induction Motors using an Acoustic Emission Sensor and Signal Processing Techniques
The 2nd FTRA International Conference on Ubiquitous Computing Application and Wireless Sensor Network (UCAWSN-14)
Abstract
This paper proposes a reliable fault diagnosis approach of induction motors using an acoustic emission (AE) sensor and signal processing techniques. In the proposed approach, discrete wavelet transform (DWT) is utilized to decompose the AE signal and select the appropriate nodes using the maximum energy ratio (MER). Then, a one-dimensional (1D) Gabor filter with various frequencies and orientation angles is applied to reduce the abnormalities and extract a number of statistical parameters. In addition, the principle component analysis (PCA) employing the extracted features is used to select the most significant feature dimensions. Finally, one against one multi-class support vector machines (OAA-MCSVMs) using the significant feature dimensions is utilized to classify each fault, where SVMs are individually trained with their own feature vectors in order to maximize the fault classification accuracy of induction motors. In the experiment, we utilize four AE fault signals which are directly collected from a R3A AE sensor attached with an induction motor. The experimental results indicate that the proposed approach achieves average classification accuracy of 99.80% and outperforms conventional fault diagnosis models in the classification accuracy.
International Journal On June-2014
Texture Feature Extraction Techniques for Fault Diagnosis of Induction Motors
Journal of Convergence
Abstract
This paper presents three texture feature extraction techniques: gray level co-occurrence matrix (GLCM), Gabor filter, and global neighborhood structure (GNS) map, for the fault diagnosis of induction motors. The texture of two- dimensional (2D) gray level images is converted from acoustic emission (AE) fault signals and used for feature extraction of the fault signals. The extracted texture features are used as inputs to a multi-class support vector machine (MCSVM) to classify each fault. The Gaussian radial basis function kernel is used with MCSVM to handle non-linear fault features of acoustic emission (AE) signals. Experimental results with one-second AE signals sampled at 1 MHz showed that the GLCM-based feature extraction method outperformed the Gabor filter and the GNS map in terms of classification accuracy because of its ability to capture the spatial dependence of gray-level texture values.
International Journal On June-2014
An Effective Approach to Improving Low-Cost GPS Positioning Accuracy in Real Time Navigation
The Science World Journal
SCIE | Impact Factor: 1.73
Abstract
Positioning accuracy is a challenging issue for location-based applications using a low-cost global positioning system (GPS). This paper presents an effective approach to improving the positioning accuracy of a low-cost GPS receiver for real-time navigation. The proposed method precisely estimates position by combining vehicle movement direction, velocity averaging, and distance between waypoints using coordinate data (latitude, longitude, time, and velocity) of the GPS receiver. The previously estimated precious reference point, coordinate translation, and invalid data check also improve accuracy. In order to evaluate the performance of the proposed method, we conducted an experiment using a GARMIN GPS 19xHVS receiver attached to a car and used Google Maps to plot the processed data. The proposed method achieved improvement of 4–10 meters in several experiments. In addition, we compared the proposed approach with two other state-of-the-art methods: recursive averaging and ARMA interpolation. The experimental results show that the proposed approach outperforms other state-of-the-art methods in terms of positioning accuracy.
Domestic Conference On June-2014
Texture Feature Analysis based Fault Diagnosis of Induction Motors
KISPS Summer Conference 2014, pp 153-157, 20-21 June 2014, Degu, South Korea.
Abstract
This paper presents three texture feature extraction techniques: gray level co-occurrence matrix (GLCM), Gabor filter, and global neighborhood structure (GNS) map, for the fault diagnosis of induction motors. The texture of two- dimensional (2D) gray level images is converted from acoustic emission (AE) fault signals and used for feature extraction of the fault signals. The extracted texture features are used as inputs to a multi-class support vector machine (MCSVM) to classify each fault. The Gaussian radial basis function kernel is used with MCSVM to handle non-linear fault features of acoustic emission (AE) signals. Experimental results with one-second AE signals sampled at 1 MHz showed that the GLCM-based feature extraction method outperformed the Gabor filter and the GNS map in terms of classification accuracy because of its ability to capture the spatial dependence of gray-level texture values.
International Conference On May-2014
Reliable RGB color image watermarking using DWT and SVD
International Conference on Informatics, Electronics & Vision (ICIEV)
Abstract
This paper proposes a reliable RGB color image watermarking which uses discrete wavelet transform (DWT) and singular value decomposition (SVD) for embedding and extracting watermark. The DWT and SVD applied on the watermark image increase information hiding capacity and perceptual similarity of the watermarked image. In the watermarking embedded stage, the processed watermark information using the proposed method is embedded into three color components (R, G and B) with an optimum watermarking scaling factor (α). In the extraction stage, the resultant watermark is calculated by averaging the three extracted watermarks from R, G and B components. The experimental results show that the proposed method achieves high normalized correlation (NC) of the extracted watermark and high peak signal to noise ratio (PSNR) of the watermarked image after several image processing attacks. In addition, the proposed method outperforms other conventional methods in terms of perceptual similarity, robustness, and detection rate.
International Conference On April-2014
Direction Averaging Method for Improving GPS Positioning Accuracy in Real-Time Navigation
The 6th FTRA International Symposium on Advances in Computing, Communications, Security, and Applications (ACSA-14)
Abstract
Positioning accuracy is a challenging issue for location-based applications using a low-cost global positioning system (GPS). This paper presents an effective approach to improving the positioning accuracy of a low-cost GPS receiver for real-time navigation. The proposed method precisely estimates position by combining vehicle movement direction, velocity averaging, and distance between waypoints using coordinate data (latitude, longitude, time, and velocity) of the GPS receiver. The previously estimated precious reference point, coordinate translation, and invalid data check also improve accuracy. In order to evaluate the performance of the proposed method, we conducted an experiment using a GARMIN GPS 19xHVS receiver attached to a car and used Google Maps to plot the processed data. The proposed method achieved improvement of 4–10 meters in several experiments. In addition, we compared the proposed approach with two other state-of-the-art methods: recursive averaging and ARMA interpolation. The experimental results show that the proposed approach outperforms other state-of-the-art methods in terms of positioning accuracy.
International Conference On April-2014
A accelerating Haegeum’s Sound Synthesis System using Graphics Processing Unit
The 6th FTRA International Symposium on Advances in Computing, Communications, Security, and Applications (ACSA-14)
Abstract
Domestic Conference On January-2014
Spectral Modeling Synthesis of Haegeum Using GPU
Winter Conference of the Korea Society of Computer and Information No. 22, Issue 1, 2014
Abstract
This paper presents a parallel approach of formant synthesis method for haegeum on graphics processing units (GPU) using spectral modeling. Spectral modeling synthesis (SMS) is a technique that models time-varying spectra as a combination of sinusoids and a time-varying filtered noise component. A second-order digital resonator by the impulse-invariant transform (IIT) is applied to generate deterministic components and the results are band-pass filtered to adjust magnitude. The noise is calculated by first generating the sinusoids with formant synthesis, subtracting them from the original sound, and then removing some harmonics remained. The synthesized sounds are consequently by adding sinusoids, which are shown to be similar to the original Haegeum sounds. Furthermore, GPU accelerates the synthesis process enabling-real time music synthesis system development, supporting more sound effect, and multiple musical sound compositions.
Domestic Conference On December-2013
Direction Averaging Method for Improving GPS Positioning Accuracy in Real Time Navigation
2013 Korea Engineering Arts Society Conference Vol. 11 no. 1, 2013
Abstract
International Journal On October-2012
Experimental Evaluation of parallelism in real time execution
Research Notes in Information Science
Abstract
International Journal On December-2011
Mobile Application and Its Global Impact
International Journal of Engineering & Technology
SCOPUS
Abstract
This paper presents the uses and effect of mobile application in individuals, business and social area. In modern information and communication age mobile application is one of the most concerned and rapidly developing areas. This paper demonstrates that how individual mobile user facilitate using mobile application and the popularity of the mobile application. Here we are presenting the consequence of mobile application in business sector. Different statistical data of the past and present situation of mobile application from different parts of the world has been presented here to express the impact. This paper also presents some effect of mobile application on society from the ethical perspective.
International Journal On April-2011
Determination of the Traveling Speed of a Moving Object of a Video Using Background Extraction and Region Based Segmentation
International Journal of Computer Science and Information Security(IJCSIS)
Abstract
This paper is concerned with the determination of thetraveling speed of a moving object of a video clip based onsubsequent object detection techniques. After preprocessing of the original image sequence, which is sampled from the videocamera, the target moving object is detected with the improvedalgorithm in which the moving object region can be extractedcompletely through several processing of background extractionand region based segmentation such as region-connection, region-merging, and region-clustering methods. Among the multiplemoving objects of the video, the target object has been detectedbased on particular criteria of region that it occupies. Then theresults of these processing can be used to determine the travelingspeed of the target moving object from changes of its coordinateposition from the video frames. Among the different video fileformat, Audio Video Interleaved (AVI) format has been used toexamine our experiments
International Journal On March-2011
Experiences and comparison study of EPC & UML for Business Process & IS Modeling
International Journal of Computer Science and Information Security (IJCSIS)
Abstract
Business process modeling is an approach by which wecan analyze and integrate the business process. Using theBusiness Process Modeling we can represent the current andfuture process of a business/organization/enterprise. The businessprocess modeling is a prerequisite and essential implementing abusiness or making any automation system. In this paper, wepresent our experience in a Business Process Modeling fororganization. This paper presents detailed description aboutbusiness process modeling, details description about the main twomodeling language EPC and UML. This paper presented theuses, advantages, disadvantages of EPC and UML modelinglanguage. Here we tried to express the experience about thosemodeling language. This paper presents a details comparisonbetween two modeling language from the business processmodeling and information system implementation point of view.
International Journal On January-2011
Organizational improvement using Organizational paradigms with the support of people paradigms
International Journal on Computer Science and Engineering (IJCSE)
Abstract
An organization is a vital part of social environment. Different parts of organization have great impact to the environment. On the other hand the different organizational strategy helps to improve the efficiency of organization and customer satisfaction. The people and tools of organization help to organization to work properly. This paper mainly describes about the organizational paradigms and people paradigms also the way how the people paradigms facilitate the organizational paradigms to improve the organizational architecture for better performance. This paper describes the different aspects of organizational like Information system strategy, Information system planning, Business process reengineering etc also End user computing, Knowledge management, Expert system of people paradigms. And finally there is a combination between those.
International Conference On January-2009
Bangla Hand Written Digit Recognition Using Supervised Locally Linear Embedding Algorithm and Support Vector Machine
12th International Conference on Computers and Information Technology, 2009. ICCIT 09.
Abstract
This paper presents Bangla numeral character recognition system using supervised locally linear embedding algorithm and support vector machine (SVM). The locally linear embedding (LLE) algorithm is an unsupervised technique proposed for nonlinear dimensionality reduction. In this paper, we describe its supervised variant (SLLE). Where class membership information is used to map overlapping high dimensional data into disjoint clusters in the embedded space. we combined it with support vector machine (SVM) for classifying handwritten digits from the on-line handwritten Bangla numeral database.
International Conference On January-2009
EMG Signal Decomposition Using Wavelet Transformation with Respect to Different Wavelet and a Comparative Study
2nd International Conference on Interaction Sciences: Information Technology, Culture and Human, Korea, ACM New York, NY, USA ©2009
Abstract
Electromyography (EMG) signals can be used for clinical/biomedical applications, Evolvable Hardware Chip (EHW) development and modern Human Computer Interaction. EMG signals acquired from muscles require advanced methods for detection, decomposition, processing, and classification. For properly analyze EMG signal there need good quality of decomposition so that it can reveal the total characteristics of EMG signals. Because EMG signal is Non-Stationary signal so it needs such a method that can decompose non-stationary signal thus wavelet decomposition is a good choice for this type. There are different types of wavelet available. Henceforth, it is necessary that proper attempt should be taken to choice the best one. Here analyses of EMG Signals were made by Various Wavelet Decomposition method with different types of wavelet and it illustrates the comparative study on best possible energy localization in the time-scale plane in order to show the performance. Thus we can choice the right one. The EMG signals used for this analysis - were found both from locally collected as well as from www.emglab.net[5] which provides EMG signal related raw data and other facilities.
International Conference On January-2009
Speaker Identification System Using PCA & Eigenface
12th International Conference on Computers and Information Technology, 2009. ICCIT 09
Abstract
This paper presents a speech-based speaker identification system and an efficient approach for selection of acoustic parameters closely related to the vocal track shape of the speaker. Speech endpoint detection algorithm is developed in order to discard the room noise and non-speech signal to achieve high accuracy of the system. Windowing and fast Fourier transform (FFT) are used to determine the spectrum of the speech signal and PCA has been used to extract feature of speech of individual speaker. Eigenface algorithm has been used here as a classification and recognition tool. Eigenspace of individual speaker is generated by the feature of the speech signal. The experimental results show the noticeable performance of the proposed system.
International Conference On January-2006
Speaker Identification System Using Eigenface Classification Engine
9th International Conference of Computer and Information Technology (ICCIT 2006)
Abstract
This paper presents a speech-based speaker identification system an efficient approach for selection of acoustic parameters closely related to the vocal track shape of the speaker. Speech endpoint detection algorithm is developed in order to discard the room noise and non-speech signal to achieve high accuracy of the system. Windowing and fast Fourier transform (FFT) are used to determine the spectrum of the speech signal of individual speaker. Eigenface algorithm is used here as the classification and identification tool. Eigenspace of individual speaker is generated by the spectrum of the speech signal. The experimental results show the noticeable performance of the proposed system.Invited Talk and Keynote Speech
-
2023The Impact of Artificial Intelligence in Industrial Applications
Keynote Speech in International Seminar on the Latest Research Trends in Technology, Comilla University, BangladeshKeynote Speech in International Seminar on the Latest Research Trends in Technology, Comilla University, Bangladesh -
2021Reliable Bearing Fault Diagnosis Systems: Introduction of Bearing Signal Processing, Feature Selection and Machine Learning Techniques
Keynote Speech in Webinar at University of Ulsan, South KoreaKeynote Speech in Webinar at University of Ulsan, South Korea . -
2021Research Process and Paper Writing- A guide for beginner
Keynote Speech in Webinar at University of Asia Pacific, BangladeshKeynote Speech in Webinar at University of Asia Pacific, Bangladesh Language: VB . -
2021Research Methodology and Research Paper Writing
Keynote Speech in Webinar at Bangladesh University of Business and Technology, BangladeshInvited Speech in Webinar at Bangladesh University of Business and Technology, Bangladesh -
2019Machine Learning Applications in HCI and Health informatics
Keynote Speech in Symposium on intelligent Information Processing at RU, BangladeshKeynote Speech in Symposium on intelligent Information Processing at RU, Bangladesh -
2019Machine Learning applications Based on Image and Signal Processing
Invited talk at BRAC University, BangladeshInvited talk at BRAC University, Bangladesh
Awards & Achievements
Honors, Awards and Grants
-
Nov-2018Best Paper Award, 3rd International Conference on Applications in Information Technology (ICAIT"18), 2018
-
2014Best Paper award, Winter Conference of the Korea Society of Computer and Information, South Korea
-
2014Best Paper award, KISPS Summer Conference 2014, South Korea
-
2014-2016BK21 Scholarship, University Ulsan, South Korea, 2014-2016
-
2013-2016Full Scholarship including tuition fees, living expenses for pursuing Ph.D. in Computing Engineering, University Ulsan, South Korea, 2013-2016
-
2010-2011Swedish Govt. awards (full tuition fees) for MSc. in Informatics at Bors (Hgskolan i Bors), Sweden, 2010-2011
-
20063rd Position, Software and Project Exhibition 2006, Dept. of CSE, University of Rajshai, Bangladesh.
-
2000-2014Merit Scholarship from University of Rajshai, 2000-2004
-
1996-1997Merit Scholarship of Bangladesh Junior Level Scholarship Exam, 1996-1997
Projects Development
Academic Projects
-
2006Vehicle Tracking System Using GPSExhibited in BASIS Softexpo 2008, Dhaka, Bangladesh (Funded By IT company, Japan)
Language: VB 6. -
2005Speaker (Human voice) Identification System (AI Based)Exhibited in IISP at BASIS Softexpo 2005, Dhaka, Bangladesh
Language: MATLAB . -
2004Simulation of 8085 Assembly Language Programming Kit.3rd place, Software and Project Exhibition 2006, CSE, RU, Rajshahi.
And Exhibit in AABISHKARER KHOJE at BASIS Softexpo2007, Dhaka
Language: VB . -
2004Web Based ticket booking and reservation SystemLanguage: PHP,
Database: MySQL. -
2003PC Based Home electronic device controllerLanguage: Visual C++.
Commercial Web application project management and development (Selected)
-
2018SiliconBasket.com: Online Superstore in Bangladesh (System Architect and team leader)
-
2017Foreverbeaute.com: Perfume E-commerce website in USA. (Team leader)
-
2016Feedbelly.com: Country wide online ordering system for restaurant in UK (System Architect and team leader)
-
2010Cuminclub.se: Online Ordering System for Restaurant, Sweden.
-
2010Restaurangmasahiro.se: Online Ordering System for Restaurant, Sweden.
-
2007-2012Rajputh.co.uk, Currymerchants.co.uk, Purbobaghrestaurant.com, spicehousestroud.co.uk, and more: Online Ordering System for Restaurant, UK.
Commercial Software project management and development (Selected)
-
2018Restaurant Point of Sell with multiple counter and APPs integrationTools: C#.Net, MySQL, Android Apps development.
Client: Orbit ITech Ltd. UK. -
2012Perfume wholesale Business Automation with AccountingLanguage: Visual C++.
Client: B & R Perfume, USA. -
2012OMR based teachers evaluation systemLanguage: C#.Net and MySQL.
Client: University of Asia Pacific, Bangladesh. -
2011Supermarket EPOS systemLanguage: VB6 and SQL Server.
Client: Costless Ltd. In Ireland. -
2009University Automation (Admission, Exam, and Accounts Management Software of Leading University, Client Server Based)Language: VB6 and SQL Server.
Client: Leading University, Bangladesh. -
2008Examination Script Management SystemLanguage: VB6 and SQL Server.
Client: Board of Intermediate and Secondary education, Sylhet (To Manage SSC and HSC exam, Script collection form center All teacher database, automatic Examiner selection, Script Distribution, Script Collection, Examiner Payment etc.) -
2007Smart ShopLanguage: VB6 and SQL Server.
Client:
1) Maha, Nahar Tower, Nayasarak, Sylhet, Bangladesh
2) Take N Pay, Mirbox Tula, Sylhet, Bangladesh
3) Top in Town, Mirbox Tula, Sylhet, Bangladesh
4) Mumme, Nahar Tower, Nayasarak, Sylhet, Bangladesh
5) Mohona Shopping center, Biany Bazar, Sylhet.
6) Samana Sopping center, Blue water, Zindabazar, Sylhet
7) Signature, Kumor para, Sylhet
8) Golden Man, Zindabazar, Sylhet. -
2007School Management SoftwareLanguage: VB6 and SQL Server.
Client:
1) Sunny Hill International School & College, Sylhet,Bangladesh.
2) Shahjalal Jamia School & College, Mira Bazar, Sylhet, Bangladesh -
2005Cold Store Management SoftwareLanguage: VB6 and SQL Server.
Client:
1) Biswash Cold Store, Nator, Bangladesh
2) Uttara Cold Store, Bogra, Bangladesh
My Photos
-
Loading photos...
Contact & Meet Me
Office Address:
[Hokkaido] 〒001-0021 Sapporo Branch- Office 305, Hokkaido University Business Spring, West 12th Street, kitaku kita 21 jō, Sapporo, Hokkaido, Japan
[Tokyo] 〒103-0007 Tokyo Branch - Office 301, TKM Nihonbashi Tower, 1-10, Nihonbashi Hamacho 2-chome, Chuo-ku, Tokyo, Japan
[Bangladesh] AI SAMURAI JAPAN Limited （Bangladesh subsidiary） － South Breeze Center (9th Floor), Building N.5, Road No.11, Block-G, Banani, Dhaka-1213
- office: +81 011-717-7017
- Mobile: +81 80 2140 7812 (JP), +88 01712501772 (BD)
- rashed.cse [at] gmail . com
- mdrashedul.islam [at] chowagiken. co. jp
- rashed.cse [at] uap-bd. edu
- rashed.cse