Rabinder Kumar Prasad
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Rabinder Kumar Prasad has a total of 19 co-authors. Here are the top co-authors by number of shared publications:
- Moirangthem Tiken Singh
Dibrugarh University
5 shared publications - Rosy Sarmah
Department of CSE, Tezpur University, Tezpur, India
5 shared publications - Abhijit Boruah
DUIET, Dibrugarh University
3 shared publications - Subrata Chakraborty
School of Science and Technology, Faculty of Science, Agriculture, Business and Law, University of New England, Armidale, NSW 2351, Australia
3 shared publications - Gurumayum Robert Michael2 shared publications
- Hemarjit Ningombam
Dibrugarh University
2 shared publications
Latest Publications
- Open Access
- 29 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 1758-1779
M k -Means: Improving Partition-Based Clustering With Optimized Centroid Initialization
The k-means algorithm is a well-known technique in partition-based clustering, drawing plenty of attention by researchers over the years. It identifies distinct, non-overlapping patterns by assigning each data point to a specific cluster using a minimum squared distance approach. However, its’ effectiveness heavily depends on the selection of initial centroid values, which are chosen randomly. Despite numerous attempts by different researchers to address this limitation, there remains significant room for improvement in the effective initialization of centroids. Here, we introduce a modified k-means, that is,Mk-means method to precisely detect all clusters in a given dataset by utilizing as objective function referred to as the sum of squared errors. This method utilizes the partial sum of the squared errors of each cluster to improve cluster identification. The efficacy of the Mk-means method is evaluated using 15 synthetic benchmark datasets and 15 real-world benchmark datasets obtained from the UCI repository. Performance comparisons were conducted against several clustering algorithms, including the traditional k-means, FCM, k-means++, and more recent approaches such as IKMN, t-k-means, fast-t-k-means, fast-t-k-means++, and Global-k-means++. The results demonstrate that Mk-means outperforms these methods in terms of external validity indices- Adjusted Rand Index (ARI), Normalized Mutual Information (NMI), and Purity, as well as sum of squared error (SSE), an internal validity index. These findings highlight that Mk-means not only achieves more accurate clustering but also reduces within-cluster variance compared to its counterparts. The source code for the Mk-means algorithm is publicly available.
- 29 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 1758-1779
Open AccessM k -Means: Improving Partition-Based Clustering With Optimized Centroid Initialization
The k-means algorithm is a well-known technique in partition-based clustering, drawing plenty of attention by researchers over the years. It identifies distinct, non-overlapping patterns by assigning each data point to a specific cluster using a minimum squared distance approach. However, its’ effectiveness heavily depends on the selection of initial centroid values, which are chosen randomly. Despite numerous attempts by different researchers to address this limitation, there remains significant room for improvement in the effective initialization of centroids. Here, we introduce a modified k-means, that is,Mk-means method to precisely detect all clusters in a given dataset by utilizing as objective function referred to as the sum of squared errors. This method utilizes the partial sum of the squared errors of each cluster to improve cluster identification. The efficacy of the Mk-means method is evaluated using 15 synthetic benchmark datasets and 15 real-world benchmark datasets obtained from the UCI repository. Performance comparisons were conducted against several clustering algorithms, including the traditional k-means, FCM, k-means++, and more recent approaches such as IKMN, t-k-means, fast-t-k-means, fast-t-k-means++, and Global-k-means++. The results demonstrate that Mk-means outperforms these methods in terms of external validity indices- Adjusted Rand Index (ARI), Normalized Mutual Information (NMI), and Purity, as well as sum of squared error (SSE), an internal validity index. These findings highlight that Mk-means not only achieves more accurate clustering but also reduces within-cluster variance compared to its counterparts. The source code for the Mk-means algorithm is publicly available.
- Open Access
- 4 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 61299-61311
Optimizing Fairness and Spectral Efficiency With Shapley-Based User Prioritization in Semantic Communication
Efficient and fair resource allocation is a critical challenge in modern semantic communication systems, particularly in scenarios where numerous users compete for limited resources. Traditional methods, such as the Hungarian algorithm and greedy allocation, primarily focus on maximizing spectral efficiency, often neglecting fairness in resource distribution. This paper introduces a Shapley value-based framework that integrates Semantic Spectral Efficiency (S-SE) with fairness-driven prioritization. By leveraging Shapley values, the framework quantifies each user’s marginal contribution to overall system performance, enabling equitable and context-aware resource allocation. Semantic similarity is embedded directly into the allocation process, allowing the proposed method to intelligently manage resources based on channel conditions and the semantic relevance of transmitted information. Experimental evaluations demonstrate the framework’s effectiveness, achieving significant improvements in fairness, as measured by Jain’s Fairness Index, without compromising spectral efficiency. The Shapley-based approach outperforms established methods, including the Hungarian algorithm, reinforcement learning algorithms like Deep Q-Network (DQN) and Proximal Policy Optimization (PPO), as well as conventional 4G and 5G resource allocation strategies. Notably, as the number of channels increases, S-SE stabilizes while fairness continues to improve, approaching optimal levels in diverse system configurations. Although the Shapley value calculation introduces a moderate increase in computational cost, this trade-off is justified by the robust balance achieved between fairness, efficiency, and overall system performance. These results highlight the potential of the proposed framework to advance next-generation wireless communication systems by prioritizing semantic relevance and equitable resource allocation.
- 4 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 61299-61311
Open AccessOptimizing Fairness and Spectral Efficiency With Shapley-Based User Prioritization in Semantic Communication
Efficient and fair resource allocation is a critical challenge in modern semantic communication systems, particularly in scenarios where numerous users compete for limited resources. Traditional methods, such as the Hungarian algorithm and greedy allocation, primarily focus on maximizing spectral efficiency, often neglecting fairness in resource distribution. This paper introduces a Shapley value-based framework that integrates Semantic Spectral Efficiency (S-SE) with fairness-driven prioritization. By leveraging Shapley values, the framework quantifies each user’s marginal contribution to overall system performance, enabling equitable and context-aware resource allocation. Semantic similarity is embedded directly into the allocation process, allowing the proposed method to intelligently manage resources based on channel conditions and the semantic relevance of transmitted information. Experimental evaluations demonstrate the framework’s effectiveness, achieving significant improvements in fairness, as measured by Jain’s Fairness Index, without compromising spectral efficiency. The Shapley-based approach outperforms established methods, including the Hungarian algorithm, reinforcement learning algorithms like Deep Q-Network (DQN) and Proximal Policy Optimization (PPO), as well as conventional 4G and 5G resource allocation strategies. Notably, as the number of channels increases, S-SE stabilizes while fairness continues to improve, approaching optimal levels in diverse system configurations. Although the Shapley value calculation introduces a moderate increase in computational cost, this trade-off is justified by the robust balance achieved between fairness, efficiency, and overall system performance. These results highlight the potential of the proposed framework to advance next-generation wireless communication systems by prioritizing semantic relevance and equitable resource allocation.
- Open Access
Spatial-Temporal Bearing Fault Detection Using Graph Attention Networks and LSTM
Purpose: This paper aims to enhance bearing fault diagnosis in industrial machinery by introducing a novel method that combines Graph Attention Network (GAT) and Long Short-Term Memory (LSTM) networks. This approach captures both spatial and temporal dependencies within sensor data, improving the accuracy of bearing fault detection under various conditions. Methodology: The proposed method converts time series sensor data into graph representations. GAT captures spatial relationships between components, while LSTM models temporal patterns. The model is validated using the Case Western Reserve University (CWRU) Bearing Dataset, which includes data under different horsepower levels and both normal and faulty conditions. Its performance is compared with methods such as K-Nearest Neighbors (KNN), Local Outlier Factor (LOF), Isolation Forest (IForest) and GNN-based method for bearing fault detection (GNNBFD). Findings: The model achieved outstanding results, with precision, recall, and F1-scores reaching 100\% across various testing conditions. It not only identifies faults accurately but also generalizes effectively across different operational scenarios, outperforming traditional methods. Originality: This research presents a unique combination of GAT and LSTM for fault detection, overcoming the limitations of traditional time series methods by capturing complex spatial-temporal dependencies. Its superior performance demonstrates significant potential for predictive maintenance in industrial applications.
Open AccessSpatial-Temporal Bearing Fault Detection Using Graph Attention Networks and LSTM
Purpose: This paper aims to enhance bearing fault diagnosis in industrial machinery by introducing a novel method that combines Graph Attention Network (GAT) and Long Short-Term Memory (LSTM) networks. This approach captures both spatial and temporal dependencies within sensor data, improving the accuracy of bearing fault detection under various conditions. Methodology: The proposed method converts time series sensor data into graph representations. GAT captures spatial relationships between components, while LSTM models temporal patterns. The model is validated using the Case Western Reserve University (CWRU) Bearing Dataset, which includes data under different horsepower levels and both normal and faulty conditions. Its performance is compared with methods such as K-Nearest Neighbors (KNN), Local Outlier Factor (LOF), Isolation Forest (IForest) and GNN-based method for bearing fault detection (GNNBFD). Findings: The model achieved outstanding results, with precision, recall, and F1-scores reaching 100\% across various testing conditions. It not only identifies faults accurately but also generalizes effectively across different operational scenarios, outperforming traditional methods. Originality: This research presents a unique combination of GAT and LSTM for fault detection, overcoming the limitations of traditional time series methods by capturing complex spatial-temporal dependencies. Its superior performance demonstrates significant potential for predictive maintenance in industrial applications.
- Open Access
Heterogeneous Graph Auto-Encoder for CreditCard Fraud Detection
The digital revolution has significantly impacted financial transactions, leading to a notable increase in credit card usage. However, this convenience comes with a trade-off: a substantial rise in fraudulent activities. Traditional machine learning methods for fraud detection often struggle to capture the inherent interconnectedness within financial data. This paper proposes a novel approach for credit card fraud detection that leverages Graph Neural Networks (GNNs) with attention mechanisms applied to heterogeneous graph representations of financial data. Unlike homogeneous graphs, heterogeneous graphs capture intricate relationships between various entities in the financial ecosystem, such as cardholders, merchants, and transactions, providing a richer and more comprehensive data representation for fraud analysis. To address the inherent class imbalance in fraud data, where genuine transactions significantly outnumber fraudulent ones, the proposed approach integrates an autoencoder. This autoencoder, trained on genuine transactions, learns a latent representation and flags deviations during reconstruction as potential fraud. This research investigates two key questions: (1) How effectively can a GNN with an attention mechanism detect and prevent credit card fraud when applied to a heterogeneous graph? (2) How does the efficacy of the autoencoder with attention approach compare to traditional methods? The results are promising, demonstrating that the proposed model outperforms benchmark algorithms such as Graph Sage and FI-GRL, achieving a superior AUC-PR of 0.89 and an F1-score of 0.81. This research significantly advances fraud detection systems and the overall security of financial transactions by leveraging GNNs with attention mechanisms and addressing class imbalance through an autoencoder.
Open AccessHeterogeneous Graph Auto-Encoder for CreditCard Fraud Detection
The digital revolution has significantly impacted financial transactions, leading to a notable increase in credit card usage. However, this convenience comes with a trade-off: a substantial rise in fraudulent activities. Traditional machine learning methods for fraud detection often struggle to capture the inherent interconnectedness within financial data. This paper proposes a novel approach for credit card fraud detection that leverages Graph Neural Networks (GNNs) with attention mechanisms applied to heterogeneous graph representations of financial data. Unlike homogeneous graphs, heterogeneous graphs capture intricate relationships between various entities in the financial ecosystem, such as cardholders, merchants, and transactions, providing a richer and more comprehensive data representation for fraud analysis. To address the inherent class imbalance in fraud data, where genuine transactions significantly outnumber fraudulent ones, the proposed approach integrates an autoencoder. This autoencoder, trained on genuine transactions, learns a latent representation and flags deviations during reconstruction as potential fraud. This research investigates two key questions: (1) How effectively can a GNN with an attention mechanism detect and prevent credit card fraud when applied to a heterogeneous graph? (2) How does the efficacy of the autoencoder with attention approach compare to traditional methods? The results are promising, demonstrating that the proposed model outperforms benchmark algorithms such as Graph Sage and FI-GRL, achieving a superior AUC-PR of 0.89 and an F1-score of 0.81. This research significantly advances fraud detection systems and the overall security of financial transactions by leveraging GNNs with attention mechanisms and addressing class imbalance through an autoencoder.
- Open Access
- 31 July 2023
- Published by ACCENTS in International Journal of Advanced Technology and Engineering Exploration
- Vol. 10 (104)
Investigating the performance vulnerability of AODV protocol of IoT network under SYN-flood attack
- 31 July 2023
- Published by ACCENTS in International Journal of Advanced Technology and Engineering Exploration
- Vol. 10 (104)
Open AccessInvestigating the performance vulnerability of AODV protocol of IoT network under SYN-flood attack
- 28 April 2023
- Published by Elsevier in Expert Systems with Applications
- Vol. 227
NNVDC: A new versatile density-based clustering method using k-Nearest Neighbors
- 28 April 2023
- Published by Elsevier in Expert Systems with Applications
- Vol. 227
NNVDC: A new versatile density-based clustering method using k-Nearest Neighbors
- 17 February 2023
- Published by Springer Nature
- p. 927-937
Real-Time Hand Gesture Recognition Using Indian Sign Language
- 17 February 2023
- Published by Springer Nature
- p. 927-937
Real-Time Hand Gesture Recognition Using Indian Sign Language
- 2 September 2022
- Published by Springer Nature in International Journal of Information Technology
- Vol. 15 (2) , 627-642
Impact of distance measures on partition-based clustering method—an empirical investigation
- 2 September 2022
- Published by Springer Nature in International Journal of Information Technology
- Vol. 15 (2) , 627-642
Impact of distance measures on partition-based clustering method—an empirical investigation
- 22 May 2022
- Published by Springer Nature
- p. 71-84
Lockdown Phases and Changing Clusters of Indian States with Respect to Number of Cases of COVID-19
- 22 May 2022
- Published by Springer Nature
- p. 71-84
Lockdown Phases and Changing Clusters of Indian States with Respect to Number of Cases of COVID-19