Rajesh Tiwari
Research Subjects
Rajesh Tiwari has a total of 49 co-authors. Here are the top co-authors by number of shared publications:
Samta Gajbhiye
Computer Science and Engineering Shri Shankarachrya Technical Campus 490020 Bhilai Chhattisgarh India
3 shared publications- Ashok Kumar
College of Computing Sciences and Information Technology, Teerthanker Mahaveer University, Moradabad, India
2 shared publications Manisha Sharma
2 shared publicationsRadhe Shyam Panda
Bhilai Institute of Technology
2 shared publications- Rajeev Shrivastava
Princeton Institute of Engineering and Technology for Women
2 shared publications - Tushar Kumar Pandey
College of Community Science, Central Agricultural University, Tura, India
2 shared publications
Latest Publications
- Open Access
- 25 October 2025
- Published by Institute for Advanced Studies
- Vol. 33 (7) , 2403-2429
A Hybrid Generative Framework for Semantic Text-to-Image Visualization Using Scene Graphs and Diffusion Models
This paper presents a new advanced framework for the AI-based text-to-image interpretation systemthat transforms unstructured natural language input into relevant, meaningful, and visually appealingimages. The system performs
- 25 October 2025
- Published by Institute for Advanced Studies
- Vol. 33 (7) , 2403-2429
Open AccessA Hybrid Generative Framework for Semantic Text-to-Image Visualization Using Scene Graphs and Diffusion Models
This paper presents a new advanced framework for the AI-based text-to-image interpretation systemthat transforms unstructured natural language input into relevant, meaningful, and visually appealingimages. The system performs
- Open Access
- 26 September 2025
- Published by Springer Nature in Research Square
Blockchain-Integrated IoT Framework for Patient Monitoring Using Optimized Multi-Scale Adaptive Graph Neural Networks
The increasing use of Internet of Things (IoT) devices in healthcare has significantly improved real-time patient monitoring. However, it also introduces critical challenges in data security, patient privacy, and the integration of heterogeneous data streams. To address these issues, we propose the first framework integrating Hyperledger Fabric with a Multi-Scale Adaptive Graph Neural Network (GNN) for real-time IoT healthcare systems , offering unmatched security and predictive accuracy. The blockchain layer ensures tamper-proof storage, enforces fine-grained access control, and builds trust among stakeholders, while the adaptive GNN captures temporal dynamics and complex interdependencies in physiological signals to detect anomalies. Experimental results on synthetic healthcare datasets show that our framework achieves a 99.6% accuracy, an AUC and F1-score of 1.0, and a blockchain throughput of 97.13 transactions per second , significantly outperforming traditional machine learning models by 10–15% in accuracy and delivering 2× higher throughput compared to prior blockchain-based systems. While current validation is based on synthetic data, real-world deployment is planned to confirm generalizability. This integrated approach paves the way for scalable, trustworthy, and equitable digital healthcare solutions , especially in resource-constrained and decentralized settings.
- 26 September 2025
- Published by Springer Nature in Research Square
Open AccessBlockchain-Integrated IoT Framework for Patient Monitoring Using Optimized Multi-Scale Adaptive Graph Neural Networks
The increasing use of Internet of Things (IoT) devices in healthcare has significantly improved real-time patient monitoring. However, it also introduces critical challenges in data security, patient privacy, and the integration of heterogeneous data streams. To address these issues, we propose the first framework integrating Hyperledger Fabric with a Multi-Scale Adaptive Graph Neural Network (GNN) for real-time IoT healthcare systems , offering unmatched security and predictive accuracy. The blockchain layer ensures tamper-proof storage, enforces fine-grained access control, and builds trust among stakeholders, while the adaptive GNN captures temporal dynamics and complex interdependencies in physiological signals to detect anomalies. Experimental results on synthetic healthcare datasets show that our framework achieves a 99.6% accuracy, an AUC and F1-score of 1.0, and a blockchain throughput of 97.13 transactions per second , significantly outperforming traditional machine learning models by 10–15% in accuracy and delivering 2× higher throughput compared to prior blockchain-based systems. While current validation is based on synthetic data, real-world deployment is planned to confirm generalizability. This integrated approach paves the way for scalable, trustworthy, and equitable digital healthcare solutions , especially in resource-constrained and decentralized settings.
- 29 August 2025
- Published by Springer Nature
- p. 317-326
Comprehensive Evaluation of Machine Learning Methods for Autism Spectrum Disorder (ASD) Prediction
- 29 August 2025
- Published by Springer Nature
- p. 317-326
Comprehensive Evaluation of Machine Learning Methods for Autism Spectrum Disorder (ASD) Prediction
- 26 March 2025
- Published by Springer Nature
- p. 379-389
Employing SVM and ANN for Feature Selection in Network Intrusion Detection Analysis
- 26 March 2025
- Published by Springer Nature
- p. 379-389
Employing SVM and ANN for Feature Selection in Network Intrusion Detection Analysis
- 25 March 2025
- Published by Springer Nature
- p. 129-140
A Survey Paper on Text Visualization Using Generative Adversarial Network
- 25 March 2025
- Published by Springer Nature
- p. 129-140
A Survey Paper on Text Visualization Using Generative Adversarial Network
- 28 February 2025
- Published by Springer Nature
- p. 337-352
Novel CBIR System for Color Logo Image Retrieval by Feature Fusion Technique
- 28 February 2025
- Published by Springer Nature
- p. 337-352
Novel CBIR System for Color Logo Image Retrieval by Feature Fusion Technique
- 1 January 2025
- Published by American Scientific Publishing Group (ASPG) LLC in Journal of Cybersecurity and Information Management
- Vol. 15 (1) , 277-77-287
Biometrics Applied to Forensics Exploring New Frontiers in Criminal Identification
American scientific publishing group
- 1 January 2025
- Published by American Scientific Publishing Group (ASPG) LLC in Journal of Cybersecurity and Information Management
- Vol. 15 (1) , 277-77-287
Biometrics Applied to Forensics Exploring New Frontiers in Criminal Identification
American scientific publishing group
- Open Access
- 1 November 2024
- Published by Institute of Advanced Engineering and Science in Indonesian Journal of Electrical Engineering and Computer Science
- Vol. 36 (2) , 1253-1260
Parameter tuning for enhancing performance of a variant of particle swarm optimization algorithm
There is dependably an extraordinary requirement for new types of algorithms in the population-based improvement algorithm. These algorithms improve the execution of the current algorithm. Parameter change approach assumes an essential job in improving the execution of the PSO algorithm. A new algorithm called particle acceleration-based particle swarm optimization (PA-PSO) has been proposed. In this algorithm a particle acceleration parameter is tuned. This algorithm significantly improves the performance of the PSO–time varying acceleration coefficients (PSO-TVAC) algorithm. This algorithm reduces the time varying weight of inertia and the nonlinear acceleration coefficients in the equation of the PSO-TVAC velocity vector in each iteration. Particle movements in the n-dimensional search space are governed by the kinetics of the second motion equation. Experiments demonstrate that the proposed PA-PSO algorithm outperforms the existing PSO-TVAC algorithm on five well-known reference test functions. The algorithm possesses adequate control over the local as well as global optimums.
- 1 November 2024
- Published by Institute of Advanced Engineering and Science in Indonesian Journal of Electrical Engineering and Computer Science
- Vol. 36 (2) , 1253-1260
Open AccessParameter tuning for enhancing performance of a variant of particle swarm optimization algorithm
There is dependably an extraordinary requirement for new types of algorithms in the population-based improvement algorithm. These algorithms improve the execution of the current algorithm. Parameter change approach assumes an essential job in improving the execution of the PSO algorithm. A new algorithm called particle acceleration-based particle swarm optimization (PA-PSO) has been proposed. In this algorithm a particle acceleration parameter is tuned. This algorithm significantly improves the performance of the PSO–time varying acceleration coefficients (PSO-TVAC) algorithm. This algorithm reduces the time varying weight of inertia and the nonlinear acceleration coefficients in the equation of the PSO-TVAC velocity vector in each iteration. Particle movements in the n-dimensional search space are governed by the kinetics of the second motion equation. Experiments demonstrate that the proposed PA-PSO algorithm outperforms the existing PSO-TVAC algorithm on five well-known reference test functions. The algorithm possesses adequate control over the local as well as global optimums.
- 20 October 2024
- Published by Springer Nature
- p. 1201-1209
A Comparative Analysis of Techniques Used for Software Fault Identification: A Survey
- 20 October 2024
- Published by Springer Nature
- p. 1201-1209
A Comparative Analysis of Techniques Used for Software Fault Identification: A Survey
- 20 October 2024
- Published by Springer Nature
- p. 1187-1200
The Efficient Link Prediction Technique for Energy Conservation of Community in Wireless Sensor Network
- 20 October 2024
- Published by Springer Nature
- p. 1187-1200
The Efficient Link Prediction Technique for Energy Conservation of Community in Wireless Sensor Network