Atul Kumar
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Atul Kumar has a total of 44 co-authors. Here are the top co-authors by number of shared publications:
- Uma Shankar Yadav
Postdoctoral Researcher, Lincoln University College, 47301, Petaling Jaya, Selangor Darul Ehsan, MALAYSIA
6 shared publications - Ashish Baiswar
Chandigarh University Uttar Pradesh
5 shared publications - Mahendra Kumar
Chandigarh University Uttar Pradesh
3 shared publications - Ajay Kumar Yadav
Registrar Chandigarh University Uttar Pradesh, Unnao, Uttar Pradesh, INDIA
2 shared publications - Ashutosh Bajpai
Chandigarh University Uttar Pradesh
2 shared publications - Dileep Kumar
Chandigarh University Uttar Pradesh
2 shared publications
Latest Publications
- 2 August 2026
- Published by Springer Nature
- p. 524-537
Deep Learning-Based Detection of Pulmonary Lung Cancer in Chest X-Rays: Clinical Application and Performance Assessment
- 2 August 2026
- Published by Springer Nature
- p. 524-537
Deep Learning-Based Detection of Pulmonary Lung Cancer in Chest X-Rays: Clinical Application and Performance Assessment
- Open Access
- 1 July 2026
- Published by Modestum Ltd in European Journal of Sustainable Development Research
- Vol. 10 (3) , em0397
Advancing sustainable development goals through multilingual text summarization: A transformer-based approach
This study uses a multilingual summarization model that is based on transformers to justify the United Nations (UN) sustainable development goals (SDGs). The multilingual and immense size of the information related to sustainability renders the synthesis of the related insights challenging policymakers, researchers, and NGOs. To handle this difficulty, we reconfigured a multilingual pre-trained transformer model (mT5) on a new customized set of sustainability texts and news articles, UN reports, and NGO reports in English, Spanish, and French. The methodology flow included the dataset generation, preprocessing, fine-tuning the model, and its thorough evaluation with the help of the automated (ROUGE) and human evaluation. The fine-tuned mT5 was found to be more semantically covered and contextually relevant than the extractive baseline that had the highest ROUGE-L and ROUGE-2 by quantitative analysis, indicating that it achieved a 42% and 39% higher ROUGE-L and ROUGE-2, respectively. The summary generated using human evaluators was rated with a higher level of coherence, fluency, factual accuracy, and SDG relevance (average score 4.2/5) as compared to extractive methods (2.8/5). The outcomes steady the theory that the fine-tuning of domain-specific multilingual processes yields significant improvement in the quality and reliability of automated sustainability summaries. The suggested framework exemplifies the way multilingual natural language processing (NLP) can facilitate access to cross-lingual knowledge and speed up the process of evidence-based decision-making in favor of the SDGs.
- 1 July 2026
- Published by Modestum Ltd in European Journal of Sustainable Development Research
- Vol. 10 (3) , em0397
Open AccessAdvancing sustainable development goals through multilingual text summarization: A transformer-based approach
This study uses a multilingual summarization model that is based on transformers to justify the United Nations (UN) sustainable development goals (SDGs). The multilingual and immense size of the information related to sustainability renders the synthesis of the related insights challenging policymakers, researchers, and NGOs. To handle this difficulty, we reconfigured a multilingual pre-trained transformer model (mT5) on a new customized set of sustainability texts and news articles, UN reports, and NGO reports in English, Spanish, and French. The methodology flow included the dataset generation, preprocessing, fine-tuning the model, and its thorough evaluation with the help of the automated (ROUGE) and human evaluation. The fine-tuned mT5 was found to be more semantically covered and contextually relevant than the extractive baseline that had the highest ROUGE-L and ROUGE-2 by quantitative analysis, indicating that it achieved a 42% and 39% higher ROUGE-L and ROUGE-2, respectively. The summary generated using human evaluators was rated with a higher level of coherence, fluency, factual accuracy, and SDG relevance (average score 4.2/5) as compared to extractive methods (2.8/5). The outcomes steady the theory that the fine-tuning of domain-specific multilingual processes yields significant improvement in the quality and reliability of automated sustainability summaries. The suggested framework exemplifies the way multilingual natural language processing (NLP) can facilitate access to cross-lingual knowledge and speed up the process of evidence-based decision-making in favor of the SDGs.
- Open Access
- 14 May 2026
- Published by Springer Nature in Discover Global Society
- Vol. 4 (1)
An ML-driven decision support system for engineering college selection post-intermediate education in India
- 14 May 2026
- Published by Springer Nature in Discover Global Society
- Vol. 4 (1)
Open AccessAn ML-driven decision support system for engineering college selection post-intermediate education in India
- 1 May 2026
- Published by Springer Nature
- p. 42-54
Quantum-Resistant Blockchain Systems: A Comprehensive Review of Post-quantum Cryptography Integration
- 1 May 2026
- Published by Springer Nature
- p. 42-54
Quantum-Resistant Blockchain Systems: A Comprehensive Review of Post-quantum Cryptography Integration
- Open Access
- 24 April 2026
- Published by Elsevier in Franklin Open
- Vol. 15
MIoT-FL: A privacy-preserving federated learning architecture for securing ECG data in heart disease prediction
- 24 April 2026
- Published by Elsevier in Franklin Open
- Vol. 15
Open AccessMIoT-FL: A privacy-preserving federated learning architecture for securing ECG data in heart disease prediction
- Open Access
- 1 April 2026
- Published by Modestum Ltd in European Journal of Sustainable Development Research
- Vol. 10 (2) , em0385
Intelligent prediction of weed-borne diseases in crops: A machine learning approach
Crop pathogens frequently find alternative hosts and a reservoir in weed species, making it harder to manage the disease and causing major loss of yield. The study discusses the use of smart methods which are machine learning and computer vision to detect the transmission of weed based diseases to farm produce in early stages. We created a predictive system that combines the process of image analysis of weed health with environmental data and agronomic data. This system is based on convolutional neural networks (CNNs) to locate the visual manifestations of the disease on common weed and a random forest classifier to estimate the possibility of further infection of the crop. Field trials on major crops and the weeds associated with them were performed to obtain data on visual symptoms of the weeds, as well as disease incidence in crops. The model that was developed showed prediction accuracy of 88 percent on occurrence of high-risk conditions that result in disease transfer. These results will show that the intelligent systems will be able to issue timely alerts, and proactive and specific disease management methods can be implemented in precision agriculture. The strategy has significantly enhanced the traditional methods of scouting since it automatizes. and facilitates the detection of the threats of disease that arise with the appearance of weeds.
- 1 April 2026
- Published by Modestum Ltd in European Journal of Sustainable Development Research
- Vol. 10 (2) , em0385
Open AccessIntelligent prediction of weed-borne diseases in crops: A machine learning approach
Crop pathogens frequently find alternative hosts and a reservoir in weed species, making it harder to manage the disease and causing major loss of yield. The study discusses the use of smart methods which are machine learning and computer vision to detect the transmission of weed based diseases to farm produce in early stages. We created a predictive system that combines the process of image analysis of weed health with environmental data and agronomic data. This system is based on convolutional neural networks (CNNs) to locate the visual manifestations of the disease on common weed and a random forest classifier to estimate the possibility of further infection of the crop. Field trials on major crops and the weeds associated with them were performed to obtain data on visual symptoms of the weeds, as well as disease incidence in crops. The model that was developed showed prediction accuracy of 88 percent on occurrence of high-risk conditions that result in disease transfer. These results will show that the intelligent systems will be able to issue timely alerts, and proactive and specific disease management methods can be implemented in precision agriculture. The strategy has significantly enhanced the traditional methods of scouting since it automatizes. and facilitates the detection of the threats of disease that arise with the appearance of weeds.
- 24 February 2026
- Published by Springer Nature
- p. 55-64
A Feature-Enriched Extractive Text Summarization Framework for Hindi Language Documents
- 24 February 2026
- Published by Springer Nature
- p. 55-64
A Feature-Enriched Extractive Text Summarization Framework for Hindi Language Documents
- 2 January 2026
- Published by Springer Nature
- p. 219-229
Review on Security Schemes in Modern IoT Integrated Cloud Systems
- 2 January 2026
- Published by Springer Nature
- p. 219-229
Review on Security Schemes in Modern IoT Integrated Cloud Systems
- Open Access
- 1 January 2026
- Published by Modestum Ltd in European Journal of Sustainable Development Research
- Vol. 10 (1) , em0351
Cross-border hydro-politics and resource security in the Yarlung Tsangpo-Brahmaputra basin: Challenges and policy imperatives
Hydropolitics between China, India, and neighboring South Asian nations, especially concerning shared water resources along the Yarlung Tsangpo River (Brahmaputra in India), is a significant concern. This study examines political, economic, and environmental interactions within the basin across China, India, Bangladesh, and Bhutan. It assesses China’s hydropower projects' impacts on downstream regions, notably in India, emphasizing challenges like water availability, ecological balance, and livelihoods. While China delays its water diversion plans, midstream dam construction poses challenges. The paper advocates cooperative frameworks for effective water resource management, underscoring transboundary water issues' role in regional geopolitics. It explores broader consequences of water insecurity on socio-economic development and environmental sustainability, contributing to understanding Yarlung Tsangpo River hydro-politics amidst growing water scarcity and competition in Asia. This research highlights four major dam projects which cumulatively exceed 1500 MW in capacity and threaten more than 20 million people living downstream. The results have practical implications for developing cooperative models and pre-emptive systems for managing trans-boundary rivers in South Asia.
- 1 January 2026
- Published by Modestum Ltd in European Journal of Sustainable Development Research
- Vol. 10 (1) , em0351
Open AccessCross-border hydro-politics and resource security in the Yarlung Tsangpo-Brahmaputra basin: Challenges and policy imperatives
Hydropolitics between China, India, and neighboring South Asian nations, especially concerning shared water resources along the Yarlung Tsangpo River (Brahmaputra in India), is a significant concern. This study examines political, economic, and environmental interactions within the basin across China, India, Bangladesh, and Bhutan. It assesses China’s hydropower projects' impacts on downstream regions, notably in India, emphasizing challenges like water availability, ecological balance, and livelihoods. While China delays its water diversion plans, midstream dam construction poses challenges. The paper advocates cooperative frameworks for effective water resource management, underscoring transboundary water issues' role in regional geopolitics. It explores broader consequences of water insecurity on socio-economic development and environmental sustainability, contributing to understanding Yarlung Tsangpo River hydro-politics amidst growing water scarcity and competition in Asia. This research highlights four major dam projects which cumulatively exceed 1500 MW in capacity and threaten more than 20 million people living downstream. The results have practical implications for developing cooperative models and pre-emptive systems for managing trans-boundary rivers in South Asia.
- 10 October 2025
- Published by Springer Nature
- p. 67-77
A Secure Framework for Source Routing in Autonomous Systems
- 10 October 2025
- Published by Springer Nature
- p. 67-77
A Secure Framework for Source Routing in Autonomous Systems