Hi, I'm Sourav. 

I am a second-year Master's Student in the College of IST at Pennsylvania State University, under the guidance of Dr. Jonathan Dodge. 

I am funded by the Kitware MIXTAPE grant, where my research focuses on developing reinforcement learning agents for strategy games, with applications in wargaming scenarios. I aim to enhance AI-driven battle tactics by addressing challenges like long-term planning and multi-agent interactions in complex environments. A key aspect of my work is ensuring that these AI strategies are explainable and interpretable by enhancing transparency through the creation of XAI middleware for both centralized and distributed control systems. 

I am also exploring how reinforcement learning can be integrated with large language models (LLMs) to improve their adaptability to new tasks and environments. I aim to develop reinforcement learning methods that enable LLMs to continuously learn and refine their performance, making them more robust and versatile across various applications. 

Latest News🔥

• Two Papers have been accepted in the NeurIPS 2024 Workshop on IMOL and Red Teaming GenAI.

Publications

• [4] Sourav Panda, Aviral Srivastava, and Jonathan Dodge. ”Unlocking New Strategies: Intrinsic Exploration for Evolving Macro and Micro Actions”. (NeurIPS 2024 workshop on Intrinsically Motivated Open-Ended Learning). [Link]

• [3] Aviral Srivastava, and Sourav Panda. ”A Formal Framework for Assessing and Mitigating Emergent Security Risks in Generative AI Models: Bridging Theory and Dynamic Risk Mitigation”. (NeurIPS 2024 workshop on Red Teaming GenAI: What Can We Learn from Adversaries?). [Link]

• [2] Sujay Koujalgi, Andrew Anderson, Iyadunni Adenuga, Shikha Soneji, Rupika Dikkala, Teresita Guzman Nader, Leo Soccio, Sourav Panda, Rupak Kumar das, Margaret Burnett, and Jonathan Dodge. ”Beyond Binary: Analyzing Human-AI Prediction Accuracy with Partial Credit Metrics”. (Manuscript in preparation for ACM-TOSEM 2024). [Link]

• [1] Brian Hu, Jonathan Dodge, Abhinav Verma, Tanmay Ambadkar, Sourav Panda, Sujay Koujalgi, Aashish Chaudhary, Brianna Major, and Bryon Lewis ”MIXTAPE: Middleware for Interactive XAI with Tree-Based AI Performance Evaluation” (2024 Simulation Interoperability Standards Organization SIMposium). [Link]

Outside the lab, I enjoy maintaining an active lifestyle through gym workouts, football, biking, and swimming. 

I am also an avid dog lover, always eager to meet new furry friends!  Tell your dog I said Hii👋