Vikas Hassija
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
Vikas Hassija has a total of 84 co-authors. Here are the top co-authors by number of shared publications:
- Vinay Chamola
BITS Pilani
15 shared publications - G. S. S. Chalapathi
Birla Institute of Technology and Science (BITS-Pilani)
10 shared publications - Qaiser Razi
Birla Institute of Technology and Science (BITS-Pilani)
6 shared publications - Anushka Singh
School of Computer Engineering, Kalinga Institute of Industrial Technology (KIIT), Bhubaneswar, Odisha, India
4 shared publications - Biplab Sikdar
National University of Singapore
4 shared publications - Arjab Chakrabarti
School of Computer Engineering, Kalinga Institute of Industrial Technology (KIIT), Bhubaneswar, Odisha, India
3 shared publications
Latest Publications
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Emerging Topics in Computational Intelligence
- Vol. 10  (5) , 3668-3681
Generation and Evaluation of Privacy-Preserving Tabular Synthetic Data Using Self-Attention Generative Adversarial Networks
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Emerging Topics in Computational Intelligence
- Vol. 10  (5) , 3668-3681
Generation and Evaluation of Privacy-Preserving Tabular Synthetic Data Using Self-Attention Generative Adversarial Networks
- Open Access
- 24 August 2026
- Published by Springer Nature in Cognitive Computation
- Vol. 18  (1) , 103
From Language Models to Agentic AI: A Survey of Autonomous, Action-Enabled, and Collaborative LLM Agents
- 24 August 2026
- Published by Springer Nature in Cognitive Computation
- Vol. 18  (1) , 103
Open AccessFrom Language Models to Agentic AI: A Survey of Autonomous, Action-Enabled, and Collaborative LLM Agents
- Open Access
- 23 January 2026
- Published by Association for Computing Machinery (ACM)
- p. 144-151
Edge-Deployed Lightweight LLMs for Medical Imaging: Real-Time Dual Report Generation with Progression Aware Summarization
- 23 January 2026
- Published by Association for Computing Machinery (ACM)
- p. 144-151
Open AccessEdge-Deployed Lightweight LLMs for Medical Imaging: Real-Time Dual Report Generation with Progression Aware Summarization
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Vehicular Technology
- Vol. 7, 2887-2905
A Survey on LLM-Enabled AAV Digital Twins for Internet of Vehicles Systems
Vehicular Digital Twins (VDTs) are emerging as a core enabler for real-time monitoring, simulation, and intelligent control in Internet of Vehicles (IoV) systems. While prior research has largely focused on ground vehicles, the integration of Unmanned Aerial Vehicles (UAVs) into vehicular digital twin ecosystems remains underexplored. This paper presents a comprehensive survey on the role of Large Language Models (LLMs) in enabling UAV Digital Twins for autonomous and cooperative operations within IoV environments. By conceptualizing UAVs as cyber-physical entities continuously synchronized with their virtual counterparts, we analyze how LLMs act as cognitive engines for high-level reasoning, multimodal perception grounding, natural-language interaction, and adaptive mission planning. The survey introduces an LLM-enabled UAV Digital Twin architecture incorporating edge-cloud collaboration, real-time state mirroring, and safety-aware control. Communication workflows leveraging vehicular networking paradigms such as V2X are discussed to support low-latency synchronization and fleet-level coordination. Key application domains, including disaster response, infrastructure monitoring, and Drone-as-a-Service within IoV systems, are examined. Finally, open challenges related to real-time synchronization, computational constraints, and scalability are highlighted, along with future research directions involving TinyLLMs, federated learning, and edge-enabled digital twin platforms.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Vehicular Technology
- Vol. 7, 2887-2905
Open AccessA Survey on LLM-Enabled AAV Digital Twins for Internet of Vehicles Systems
Vehicular Digital Twins (VDTs) are emerging as a core enabler for real-time monitoring, simulation, and intelligent control in Internet of Vehicles (IoV) systems. While prior research has largely focused on ground vehicles, the integration of Unmanned Aerial Vehicles (UAVs) into vehicular digital twin ecosystems remains underexplored. This paper presents a comprehensive survey on the role of Large Language Models (LLMs) in enabling UAV Digital Twins for autonomous and cooperative operations within IoV environments. By conceptualizing UAVs as cyber-physical entities continuously synchronized with their virtual counterparts, we analyze how LLMs act as cognitive engines for high-level reasoning, multimodal perception grounding, natural-language interaction, and adaptive mission planning. The survey introduces an LLM-enabled UAV Digital Twin architecture incorporating edge-cloud collaboration, real-time state mirroring, and safety-aware control. Communication workflows leveraging vehicular networking paradigms such as V2X are discussed to support low-latency synchronization and fleet-level coordination. Key application domains, including disaster response, infrastructure monitoring, and Drone-as-a-Service within IoV systems, are examined. Finally, open challenges related to real-time synchronization, computational constraints, and scalability are highlighted, along with future research directions involving TinyLLMs, federated learning, and edge-enabled digital twin platforms.
- 22 December 2025
- Published by Elsevier in Vehicular Communications
- Vol. 58
The role of large language models (LLMs) in enhancing intelligent transportation systems: A survey
- 22 December 2025
- Published by Elsevier in Vehicular Communications
- Vol. 58
The role of large language models (LLMs) in enhancing intelligent transportation systems: A survey
- 11 December 2025
- Published by Springer Nature in Cognitive Computation
- Vol. 17  (6) , 175
Privacy-Preserving Skin Disease Detection Using One-Shot Federated Learning Approach
- 11 December 2025
- Published by Springer Nature in Cognitive Computation
- Vol. 17  (6) , 175
Privacy-Preserving Skin Disease Detection Using One-Shot Federated Learning Approach
- Open Access
- 28 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 201625-201644
Privacy-Preserving On-Screen Activity Recognition via One-Shot Federated Learning
Preserving user privacy while monitoring on-screen activity is a growing challenge in remote and decentralized work environments. In this paper, we propose a novel one-shot federated learning (FL) framework that enables secure, low-overhead, and privacy-preserving on-screen activity recognition. Conventional FL methods require multiple communication rounds between clients and a central server. In contrast, our one-shot FL approach performs model aggregation in a single round, significantly reducing communication costs and latency, making it ideal for bandwidth-constrained and resource-limited environments. Using a custom dataset of 2,300 screenshots from common applications (e.g., Coursera, YouTube, Amazon), we train Deep Convolutional Neural Network (DCNN) models such as VGG16, VGG19, and InceptionResNetV2 locally on client devices. Sensitive visual data is never shared, and the screenshots are processed and discarded post-inference, ensuring strict data confidentiality. We further enhance privacy by integrating differential privacy mechanisms. To further improve learning effectiveness under different data distributions, we evaluate two aggregation techniques, FedAvg for IID and FedProx for non-IID settings. Experimental evaluations show that our approach achieves up to 99.1% classification accuracy (MobileNet) while reducing communication overhead by over 80% compared to traditional FL. This work highlights the effectiveness and scalability of one-shot FL for secure, real-time activity tracking in privacy-sensitive applications.
- 28 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 201625-201644
Open AccessPrivacy-Preserving On-Screen Activity Recognition via One-Shot Federated Learning
Preserving user privacy while monitoring on-screen activity is a growing challenge in remote and decentralized work environments. In this paper, we propose a novel one-shot federated learning (FL) framework that enables secure, low-overhead, and privacy-preserving on-screen activity recognition. Conventional FL methods require multiple communication rounds between clients and a central server. In contrast, our one-shot FL approach performs model aggregation in a single round, significantly reducing communication costs and latency, making it ideal for bandwidth-constrained and resource-limited environments. Using a custom dataset of 2,300 screenshots from common applications (e.g., Coursera, YouTube, Amazon), we train Deep Convolutional Neural Network (DCNN) models such as VGG16, VGG19, and InceptionResNetV2 locally on client devices. Sensitive visual data is never shared, and the screenshots are processed and discarded post-inference, ensuring strict data confidentiality. We further enhance privacy by integrating differential privacy mechanisms. To further improve learning effectiveness under different data distributions, we evaluate two aggregation techniques, FedAvg for IID and FedProx for non-IID settings. Experimental evaluations show that our approach achieves up to 99.1% classification accuracy (MobileNet) while reducing communication overhead by over 80% compared to traditional FL. This work highlights the effectiveness and scalability of one-shot FL for secure, real-time activity tracking in privacy-sensitive applications.
- Open Access
- 19 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 197955-197966
Quantum-Accelerated Feature Selection for Edge-Enabled Perception in Connected and Autonomous Vehicles
The growing complexity and volume of traffic and sensory data in Connected and Autonomous Vehicles (CAVs) make it crucial to develop more effective methods for selecting and extracting features from high-dimensional images and videos. This paper proposes a novel quantum-accelerated feature selection framework utilising the Variational Quantum Eigensolver (VQE) to enhance perception tasks in CAV environments. By decomposing high-dimensional traffic images into compact patches encoded as quantum Hamiltonians, the proposed design aims to support Intelligent Transportation Systems (ITS) tasks such as object detection, vehicle classification, and traffic monitoring in CAVs.The quantum hybrid method serves as a pipeline for extracting selective features that have been optimised using gradient-aware parameter pruning and conditional principal component analysis (PCA). The implementation of this quantum hybrid approach facilitates addressing computational challenges at the edge in embedded vehicular systems, reducing data dimensionality while preserving key discriminative spatial patterns necessary for object detection and traffic monitoring. Although the quantum method currently underperforms in certain clustering and separability metrics, it demonstrates potential as a viable alternative for advanced perception in CAVs.This work lays a foundation for integrating quantum computing into edge-based autonomous vehicle perception, highlighting both opportunities and challenges with near-term quantum hardware.
- 19 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 197955-197966
Open AccessQuantum-Accelerated Feature Selection for Edge-Enabled Perception in Connected and Autonomous Vehicles
The growing complexity and volume of traffic and sensory data in Connected and Autonomous Vehicles (CAVs) make it crucial to develop more effective methods for selecting and extracting features from high-dimensional images and videos. This paper proposes a novel quantum-accelerated feature selection framework utilising the Variational Quantum Eigensolver (VQE) to enhance perception tasks in CAV environments. By decomposing high-dimensional traffic images into compact patches encoded as quantum Hamiltonians, the proposed design aims to support Intelligent Transportation Systems (ITS) tasks such as object detection, vehicle classification, and traffic monitoring in CAVs.The quantum hybrid method serves as a pipeline for extracting selective features that have been optimised using gradient-aware parameter pruning and conditional principal component analysis (PCA). The implementation of this quantum hybrid approach facilitates addressing computational challenges at the edge in embedded vehicular systems, reducing data dimensionality while preserving key discriminative spatial patterns necessary for object detection and traffic monitoring. Although the quantum method currently underperforms in certain clustering and separability metrics, it demonstrates potential as a viable alternative for advanced perception in CAVs.This work lays a foundation for integrating quantum computing into edge-based autonomous vehicle perception, highlighting both opportunities and challenges with near-term quantum hardware.
- 17 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13  (3) , 3808-3826
A Comprehensive Survey on Data Distillation: Techniques, Frameworks, and Future Directions
The increased adoption of machine learning techniques has led to exponential growth in data generation and utilization. This growth has necessitated efficient storage, processing, and utilization of this data, which presents critical challenges, particularly in resource-constrained environments such as the Internet of Things (IoT) and edge devices. Data distillation has emerged as a promising solution that reduces dataset size while preserving the essential information and optimizing computational resources. This survey provides a comprehensive analysis of data reduction techniques, covering methodologies, such as knowledge distillation (KD), core-set selection, hyperparameter optimization (HPO), and generative modeling. We further explore various data distillation learning frameworks, including performance, gradient, parameter, and distribution matching (DM), highlighting their effectiveness in different data modalities, such as images, graphs, and text. Furthermore, we examine the implications of data distillation in key areas, such as continual and federated learning, privacy preservation, security, healthcare, IoT applications, and edge computing. By enabling lightweight models with minimal computational overhead, data distillation facilitates real-time inference and decision-making on edge devices, making it highly relevant for low-power, bandwidth-limited environments. Data distillation offers numerous advantages in improving the model efficiency, reducing training costs, and enhancing privacy. However, data distillation faces numerous challenges related to scalability, computational complexity, and information retention. This survey identifies these challenges and outlines potential future research directions, providing insights for researchers seeking to leverage data distillation for scalable and efficient machine learning applications.
- 17 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13  (3) , 3808-3826
A Comprehensive Survey on Data Distillation: Techniques, Frameworks, and Future Directions
The increased adoption of machine learning techniques has led to exponential growth in data generation and utilization. This growth has necessitated efficient storage, processing, and utilization of this data, which presents critical challenges, particularly in resource-constrained environments such as the Internet of Things (IoT) and edge devices. Data distillation has emerged as a promising solution that reduces dataset size while preserving the essential information and optimizing computational resources. This survey provides a comprehensive analysis of data reduction techniques, covering methodologies, such as knowledge distillation (KD), core-set selection, hyperparameter optimization (HPO), and generative modeling. We further explore various data distillation learning frameworks, including performance, gradient, parameter, and distribution matching (DM), highlighting their effectiveness in different data modalities, such as images, graphs, and text. Furthermore, we examine the implications of data distillation in key areas, such as continual and federated learning, privacy preservation, security, healthcare, IoT applications, and edge computing. By enabling lightweight models with minimal computational overhead, data distillation facilitates real-time inference and decision-making on edge devices, making it highly relevant for low-power, bandwidth-limited environments. Data distillation offers numerous advantages in improving the model efficiency, reducing training costs, and enhancing privacy. However, data distillation faces numerous challenges related to scalability, computational complexity, and information retention. This survey identifies these challenges and outlines potential future research directions, providing insights for researchers seeking to leverage data distillation for scalable and efficient machine learning applications.
- Top Cited PapersOpen Access
- 20 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 95496-95523
Transformers for Vision: A Survey on Innovative Methods for Computer Vision
Transformers have emerged as a groundbreaking architecture in the field of computer vision, offering a compelling alternative to traditional convolutional neural networks (CNNs) by enabling the modeling of long-range dependencies and global context through self-attention mechanisms. Originally developed for natural language processing, transformers have now been successfully adapted for a wide range of vision tasks, leading to significant improvements in performance and generalization. This survey provides a comprehensive overview of the fundamental principles of transformer architectures, highlighting the core mechanisms such as self-attention, multi-head attention, and positional encoding that distinguish them from CNNs. We delve into the theoretical adaptations required to apply transformers to visual data, including image tokenization and the integration of positional embeddings. A detailed analysis of key transformer-based vision architectures such as ViT, DeiT, Swin Transformer, PVT, Twins, and CrossViT are presented, alongside their practical applications in image classification, object detection, video understanding, medical imaging, and cross-modal tasks. The paper further compares the performance of vision transformers with CNNs, examining their respective strengths, limitations, and the emergence of hybrid models. Finally, current challenges in deploying ViTs, such as computational cost, data efficiency, and interpretability, and explore recent advancements and future research directions including efficient architectures, self-supervised learning, and multimodal integration are discussed.
- 20 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 95496-95523
Top Cited PapersOpen AccessTransformers for Vision: A Survey on Innovative Methods for Computer Vision
Transformers have emerged as a groundbreaking architecture in the field of computer vision, offering a compelling alternative to traditional convolutional neural networks (CNNs) by enabling the modeling of long-range dependencies and global context through self-attention mechanisms. Originally developed for natural language processing, transformers have now been successfully adapted for a wide range of vision tasks, leading to significant improvements in performance and generalization. This survey provides a comprehensive overview of the fundamental principles of transformer architectures, highlighting the core mechanisms such as self-attention, multi-head attention, and positional encoding that distinguish them from CNNs. We delve into the theoretical adaptations required to apply transformers to visual data, including image tokenization and the integration of positional embeddings. A detailed analysis of key transformer-based vision architectures such as ViT, DeiT, Swin Transformer, PVT, Twins, and CrossViT are presented, alongside their practical applications in image classification, object detection, video understanding, medical imaging, and cross-modal tasks. The paper further compares the performance of vision transformers with CNNs, examining their respective strengths, limitations, and the emergence of hybrid models. Finally, current challenges in deploying ViTs, such as computational cost, data efficiency, and interpretability, and explore recent advancements and future research directions including efficient architectures, self-supervised learning, and multimodal integration are discussed.