Computer Science TechTalk Series
When: Wednesday, October 1, 2025
Time: 3:00 PM
Where: RB 104
Title: Enhancing Knowledge Discovery: Truth Discovery, LLM-Powered Learning, and Novel Hypothesis Generation
Presented By: Adithya Kulkarni – Assistant Professor of Computer Science
Abstract: This talk presents a roadmap from raw data to actionable knowledge through truth discovery, collaborative reasoning, and novel hypothesis generation. It demonstrates how the integration of large language models with graph structures and symbolic logic enables the development of uncertainty-aware and interpretable systems that mitigate hallucinations and enhance decision quality. The discussion includes case studies on budget-aware label aggregation in graphs, robust prompting without labeled data, graph- and logic-augmented reasoning, and long-conversation generation with dynamic topic tracking. The talk concludes with future directions toward building trustworthy, efficient, and human-in-the-loop AI systems for scientific and creative domains.
Bio: Adithya Kulkarni is an Assistant Professor of Computer Science at Ball State University and previously served as a Postdoctoral Fellow at Virginia Tech. His research focuses on graph-augmented reasoning, uncertainty-aware inference, and human-centered artificial intelligence. He has published in leading venues including KDD, UAI, ACL/EMNLP, IEEE BigData, SDM, ICDM, and INLG, with contributions spanning budget allocation for graph labeling, robust prompting, conformal prediction for structured models, and logic- and graph-based methods for enhancing LLM reasoning. He has also organized international workshops on agentic AI for science at ICLR, WWW, and AAAI. Prior to his academic career, he worked as a software engineer at Oracle and earned his Ph.D. in Computer Science from Iowa State University.
Light refreshments will be served