Jagat Sesh Challa
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Jagat Sesh Challa has a total of 56 co-authors. Here are the top co-authors by number of shared publications:
- Poonam Goyal
BITS Pilani Pilani Rajasthan India
3 shared publications - Dhruv Kumar
Birla Institute of Technology and Science Pilani, Pilani, Rajasthan, India and Indraprastha Institute of Information Technology Delhi, New Delhi, New Delhi, India
2 shared publications - Dhruv Kumar
Department of Computer Science & Information Systems Birla Institute of Technology and Science Pilani India
2 shared publications - Navneet Goyal
BITS Pilani Pilani Rajasthan India
2 shared publications Sundar Balasubramaniam
Birla Inst Technol & Sci, Adv Data Analyt & Parallel Technol Lab, Pilani 333031, India
2 shared publicationsAarti
1 shared publication
Latest Publications
- Open Access
- 2 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 19143-19156
B-CvT: A Style Transfer Framework for Content Preservation and Fairness in Skin Lesion Imaging
There exists a significant lack of diversity based on skin tone representation in medical imaging, which contributes to delayed cancer diagnoses and poor outcomes for skin cancers in people of color. Current diagnostic systems predominantly feature lighter skin tones, which hinders accurate pathology identification on darker skin. Neural style transfer has emerged as a promising technique to address this data imbalance by artificially expanding the diversity of skin tone representations in medical image datasets. However, existing methods often disrupt critical diagnostic features through artifacts or structural deformations. To address this problem, we present a branched convolutional vision transformer (B-CvT) encoder coupled with a convolutional transformer decoder to enable precise skin tone transfer. The architecture preserves pathological content by employing dedicated encoder branches to disentangle style (skin tone) from content (lesion morphology), reinforced by a Wasserstein distance-based similarity loss that enforces clean feature separation in latent space. Quantitative evaluation demonstrates superior performance, surpassing state-of-the-art style transfer methods with the best Learned Perceptual Image Patch Similarity (LPIPS) of 0.320 and Structural Similarity Index Measure (SSIM) of 0.803. When used as a data augmentation technique, the proposed method improves the fairness of the model in cancer diagnostics by reducing the Equalized Odds between different skin tones by 0.09 without compromising overall accuracy. This provides a high-fidelity image synthesis tool and a method for developing more equitable AI diagnostic models, addressing a critical gap in dermatological representation by preserving the core pathological content essential for downstream diagnostic tasks.
- 2 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 19143-19156
Open AccessB-CvT: A Style Transfer Framework for Content Preservation and Fairness in Skin Lesion Imaging
There exists a significant lack of diversity based on skin tone representation in medical imaging, which contributes to delayed cancer diagnoses and poor outcomes for skin cancers in people of color. Current diagnostic systems predominantly feature lighter skin tones, which hinders accurate pathology identification on darker skin. Neural style transfer has emerged as a promising technique to address this data imbalance by artificially expanding the diversity of skin tone representations in medical image datasets. However, existing methods often disrupt critical diagnostic features through artifacts or structural deformations. To address this problem, we present a branched convolutional vision transformer (B-CvT) encoder coupled with a convolutional transformer decoder to enable precise skin tone transfer. The architecture preserves pathological content by employing dedicated encoder branches to disentangle style (skin tone) from content (lesion morphology), reinforced by a Wasserstein distance-based similarity loss that enforces clean feature separation in latent space. Quantitative evaluation demonstrates superior performance, surpassing state-of-the-art style transfer methods with the best Learned Perceptual Image Patch Similarity (LPIPS) of 0.320 and Structural Similarity Index Measure (SSIM) of 0.803. When used as a data augmentation technique, the proposed method improves the fairness of the model in cancer diagnostics by reducing the Equalized Odds between different skin tones by 0.09 without compromising overall accuracy. This provides a high-fidelity image synthesis tool and a method for developing more equitable AI diagnostic models, addressing a critical gap in dermatological representation by preserving the core pathological content essential for downstream diagnostic tasks.
- Open Access
Beyond Single Bugs: Benchmarking Large Language Models for Multi-Vulnerability Detection
Large Language Models (LLMs) have demonstrated significant potential in automated software security, particularly in vulnerability detection. However, existing benchmarks primarily focus on isolated, single-vulnerability samples or function-level classification, failing to reflect the complexity of real-world software where multiple interacting vulnerabilities often coexist within large files. Recent studies indicate that LLMs suffer from "count bias" and "selection bias" in multi-label tasks, yet this has not been rigorously quantified in the domain of code security. In this work, we introduce a comprehensive benchmark for Multi-Vulnerability Detection across four major languages: C, C++, Python, and JavaScript. We construct a dataset of 40,000 files by systematically injecting controlled counts of vulnerabilities (1, 3, 5, and 9) into long-context code samples (7.5k-10k tokens) sourced from CodeParrot. We evaluate five state-of-the-art LLMs, including GPT-4o-mini, Llama-3.3-70B, and the Qwen-2.5 series. Our results reveal a sharp degradation in performance as vulnerability density increases. While Llama-3.3-70B achieves near-perfect F1 scores (approximately 0.97) on single-vulnerability C tasks, performance drops by up to 40% in high-density settings. Notably, Python and JavaScript show distinct failure modes compared to C/C++, with models exhibiting severe "under-counting" (Recall dropping to less than 0.30) in complex Python files.
Open AccessBeyond Single Bugs: Benchmarking Large Language Models for Multi-Vulnerability Detection
Large Language Models (LLMs) have demonstrated significant potential in automated software security, particularly in vulnerability detection. However, existing benchmarks primarily focus on isolated, single-vulnerability samples or function-level classification, failing to reflect the complexity of real-world software where multiple interacting vulnerabilities often coexist within large files. Recent studies indicate that LLMs suffer from "count bias" and "selection bias" in multi-label tasks, yet this has not been rigorously quantified in the domain of code security. In this work, we introduce a comprehensive benchmark for Multi-Vulnerability Detection across four major languages: C, C++, Python, and JavaScript. We construct a dataset of 40,000 files by systematically injecting controlled counts of vulnerabilities (1, 3, 5, and 9) into long-context code samples (7.5k-10k tokens) sourced from CodeParrot. We evaluate five state-of-the-art LLMs, including GPT-4o-mini, Llama-3.3-70B, and the Qwen-2.5 series. Our results reveal a sharp degradation in performance as vulnerability density increases. While Llama-3.3-70B achieves near-perfect F1 scores (approximately 0.97) on single-vulnerability C tasks, performance drops by up to 40% in high-density settings. Notably, Python and JavaScript show distinct failure modes compared to C/C++, with models exhibiting severe "under-counting" (Recall dropping to less than 0.30) in complex Python files.
- 30 October 2025
- Published by Elsevier in Computer Science Review
- Vol. 59
Time-sensitive data analytics: A survey of anytime techniques, applications and challenges
- 30 October 2025
- Published by Elsevier in Computer Science Review
- Vol. 59
Time-sensitive data analytics: A survey of anytime techniques, applications and challenges
- 20 September 2025
- Published by Springer Nature in Knowledge and Information Systems
- Vol. 67 (12) , 12179-12210
Toward a socially fair data equilibration approach for responsible federated learning
- 20 September 2025
- Published by Springer Nature in Knowledge and Information Systems
- Vol. 67 (12) , 12179-12210
Toward a socially fair data equilibration approach for responsible federated learning
- Open Access
Talking with Oompa Loompas: A novel framework for evaluating linguistic acquisition of LLM agents
Existing evaluation studies on linguistic competence of large language models (LLM agents) have focused primarily on vocabulary learning, morphological rule induction, syntactic generalization, pragmatic inference, and cross-linguistic transfer. However, none assess whether LLM agents can acquire a language through pattern recognition and interactive feedback, a central feature of human language acquisition. We propose a novel experimental framework in which an LLM agent is evaluated on its ability to acquire and use a newly constructed language (Tinkatongue) in conversation with a bot that understands only Tinkatongue. Our findings show that LLM agents fail to establish a conversation within 100 responses, yet they adopt distinct strategies that mirror human approaches to language learning. The results suggest a new direction for evaluation benchmarks and open pathways to model designs that learn more effectively from interactive feedback.
Open AccessTalking with Oompa Loompas: A novel framework for evaluating linguistic acquisition of LLM agents
Existing evaluation studies on linguistic competence of large language models (LLM agents) have focused primarily on vocabulary learning, morphological rule induction, syntactic generalization, pragmatic inference, and cross-linguistic transfer. However, none assess whether LLM agents can acquire a language through pattern recognition and interactive feedback, a central feature of human language acquisition. We propose a novel experimental framework in which an LLM agent is evaluated on its ability to acquire and use a newly constructed language (Tinkatongue) in conversation with a bot that understands only Tinkatongue. Our findings show that LLM agents fail to establish a conversation within 100 responses, yet they adopt distinct strategies that mirror human approaches to language learning. The results suggest a new direction for evaluation benchmarks and open pathways to model designs that learn more effectively from interactive feedback.
- Open Access
- 2 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 181-195
Rubric Is All You Need: Improving LLM-Based Code Evaluation With Question-Specific Rubrics
- 2 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 181-195
Open AccessRubric Is All You Need: Improving LLM-Based Code Evaluation With Question-Specific Rubrics
- Top Cited PapersOpen Access
- 17 March 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-14
The role of generative AI tools in shaping mechanical engineering education from an undergraduate perspective
This study evaluates the effectiveness of three leading generative AI tools-ChatGPT, Gemini, and Copilot-in undergraduate mechanical engineering education using a mixed-methods approach. The performance of these tools was assessed on 800 questions spanning seven core subjects, covering multiple-choice, numerical, and theory-based formats. While all three AI tools demonstrated strong performance in theory-based questions, they struggled with numerical problem-solving, particularly in areas requiring deep conceptual understanding and complex calculations. Among them, Copilot achieved the highest accuracy (60.38%), followed by Gemini (57.13%) and ChatGPT (46.63%). To complement these findings, a survey of 172 students and interviews with 20 participants provided insights into user experiences, challenges, and perceptions of AI in academic settings. Thematic analysis revealed concerns regarding AI’s reliability in numerical tasks and its potential impact on students’ problem-solving abilities. Based on these results, this study offers strategic recommendations for integrating AI into mechanical engineering curricula, ensuring its responsible use to enhance learning without fostering dependency. Additionally, we propose instructional strategies to help educators adapt assessment methods in the era of AI-assisted learning. These findings contribute to the broader discussion on AI’s role in engineering education and its implications for future learning methodologies.
- 17 March 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-14
Top Cited PapersOpen AccessThe role of generative AI tools in shaping mechanical engineering education from an undergraduate perspective
This study evaluates the effectiveness of three leading generative AI tools-ChatGPT, Gemini, and Copilot-in undergraduate mechanical engineering education using a mixed-methods approach. The performance of these tools was assessed on 800 questions spanning seven core subjects, covering multiple-choice, numerical, and theory-based formats. While all three AI tools demonstrated strong performance in theory-based questions, they struggled with numerical problem-solving, particularly in areas requiring deep conceptual understanding and complex calculations. Among them, Copilot achieved the highest accuracy (60.38%), followed by Gemini (57.13%) and ChatGPT (46.63%). To complement these findings, a survey of 172 students and interviews with 20 participants provided insights into user experiences, challenges, and perceptions of AI in academic settings. Thematic analysis revealed concerns regarding AI’s reliability in numerical tasks and its potential impact on students’ problem-solving abilities. Based on these results, this study offers strategic recommendations for integrating AI into mechanical engineering curricula, ensuring its responsible use to enhance learning without fostering dependency. Additionally, we propose instructional strategies to help educators adapt assessment methods in the era of AI-assisted learning. These findings contribute to the broader discussion on AI’s role in engineering education and its implications for future learning methodologies.
- Open Access
- 14 October 2024
- Published by Wiley in Journal of Computing and Information Science in Engineering
- Vol. 24 (12) , 1-17
A Review of the Applications of Machine Learning for Prediction and Analysis of Mechanical Properties and Microstructures in Additive Manufacturing
This article provides an insightful review of the recent applications of machine learning (ML) techniques in additive manufacturing (AM) for the prediction and amelioration of mechanical properties, as well as the analysis and prediction of microstructures. AM is the modern digital manufacturing technique adopted in various industrial sectors because of its salient features, such as the fabrication of geometrically complex and customized parts, the fabrication of parts with unique properties and microstructures, and the fabrication of hard-to-manufacture materials. The functioning of the AM processes is complicated. Several factors such as process parameters, defects, cooling rates, thermal histories, and machine stability have a prominent impact on AM products’ properties and microstructure. It is difficult to establish the relationship between these AM factors and the AM end product properties and microstructure. Several studies have utilized different ML techniques to optimize AM processes and predict mechanical properties and microstructure. This article discusses the applications of various ML techniques in AM to predict mechanical properties and optimization of AM processes for the amelioration of mechanical properties of end parts. Also, ML applications for segmentation, prediction, and analysis of AM-fabricated material’s microstructures and acceleration of microstructure prediction procedures are discussed in this article.
- 14 October 2024
- Published by Wiley in Journal of Computing and Information Science in Engineering
- Vol. 24 (12) , 1-17
Open AccessA Review of the Applications of Machine Learning for Prediction and Analysis of Mechanical Properties and Microstructures in Additive Manufacturing
This article provides an insightful review of the recent applications of machine learning (ML) techniques in additive manufacturing (AM) for the prediction and amelioration of mechanical properties, as well as the analysis and prediction of microstructures. AM is the modern digital manufacturing technique adopted in various industrial sectors because of its salient features, such as the fabrication of geometrically complex and customized parts, the fabrication of parts with unique properties and microstructures, and the fabrication of hard-to-manufacture materials. The functioning of the AM processes is complicated. Several factors such as process parameters, defects, cooling rates, thermal histories, and machine stability have a prominent impact on AM products’ properties and microstructure. It is difficult to establish the relationship between these AM factors and the AM end product properties and microstructure. Several studies have utilized different ML techniques to optimize AM processes and predict mechanical properties and microstructure. This article discusses the applications of various ML techniques in AM to predict mechanical properties and optimization of AM processes for the amelioration of mechanical properties of end parts. Also, ML applications for segmentation, prediction, and analysis of AM-fabricated material’s microstructures and acceleration of microstructure prediction procedures are discussed in this article.
- Open Access
- 14 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of the Computer Society
- Vol. 5, 406-417
InFER++: Real-World Indian Facial Expression Dataset
Detecting facial expressions is a challenging task in the field of computer vision. Several datasets and algorithms have been proposed over the past two decades; however, deploying them in real-world, in-the-wild scenarios hampers the overall performance. This is because the training data does not completely represent socio-cultural and ethnic diversity; the majority of the datasets consist of American and Caucasian populations. On the contrary, in a diverse and heterogeneous population distribution like the Indian subcontinent, the need for a significantly large enough dataset representing all the ethnic groups is even more critical. To address this, we present InFER++, an India-specific, multi-ethnic, real-world, in-the-wild facial expression dataset consisting of seven basic expressions. To the best of our knowledge, this is the largest India-specific facial expression dataset. Our cross-dataset analysis of RAF-DB vs InFER++ shows that models trained on RAF-DB were not generalizable to ethnic datasets like InFER++. This is because the facial expressions change with respect to ethnic and socio-cultural factors. We also present LiteXpressionNet, a lightweight deep facial expression network that outperforms many existing lightweight models with considerably fewer FLOPs and parameters. The proposed model is inspired by MobileViTv2 architecture, which utilizes GhostNetv2 blocks to increase parametrization while reducing latency and FLOP requirements. The model is trained with a novel objective function that combines early learning regularization and symmetric cross-entropy loss to mitigate human uncertainties and annotation bias in most real-world facial expression datasets.
- 14 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of the Computer Society
- Vol. 5, 406-417
Open AccessInFER++: Real-World Indian Facial Expression Dataset
Detecting facial expressions is a challenging task in the field of computer vision. Several datasets and algorithms have been proposed over the past two decades; however, deploying them in real-world, in-the-wild scenarios hampers the overall performance. This is because the training data does not completely represent socio-cultural and ethnic diversity; the majority of the datasets consist of American and Caucasian populations. On the contrary, in a diverse and heterogeneous population distribution like the Indian subcontinent, the need for a significantly large enough dataset representing all the ethnic groups is even more critical. To address this, we present InFER++, an India-specific, multi-ethnic, real-world, in-the-wild facial expression dataset consisting of seven basic expressions. To the best of our knowledge, this is the largest India-specific facial expression dataset. Our cross-dataset analysis of RAF-DB vs InFER++ shows that models trained on RAF-DB were not generalizable to ethnic datasets like InFER++. This is because the facial expressions change with respect to ethnic and socio-cultural factors. We also present LiteXpressionNet, a lightweight deep facial expression network that outperforms many existing lightweight models with considerably fewer FLOPs and parameters. The proposed model is inspired by MobileViTv2 architecture, which utilizes GhostNetv2 blocks to increase parametrization while reducing latency and FLOP requirements. The model is trained with a novel objective function that combines early learning regularization and symmetric cross-entropy loss to mitigate human uncertainties and annotation bias in most real-world facial expression datasets.
- 26 June 2024
- Published by Springer Nature in Journal of Computer Science and Technology
- Vol. 39 (3) , 610-636
A Survey and Experimental Review on Data Distribution Strategies for Parallel Spatial Clustering Algorithms
- 26 June 2024
- Published by Springer Nature in Journal of Computer Science and Technology
- Vol. 39 (3) , 610-636
A Survey and Experimental Review on Data Distribution Strategies for Parallel Spatial Clustering Algorithms