Amit Dhurandhar

Welcome To Amit Dhurandhar's Webpage..

I am originally from Pune, India. I am a principal research staff member at IBM T.J. Watson in Yorktown Heights NY. I completed my Ph.D. in the Department of Computer and Information Science and Engineering at the University of Florida (UF), Gainesville. My advisor was Dr. Alin Dobra. My primary research areas are Machine learning and Data Mining.

I admire originality and brilliance but believe that having the right attitude is more important in life. 

Whats new?

• Gave talk on XAI: From Classification to Generation at Workshop on Emerging ICT Trends & Applications (WEITA), 2025.
• Paper on GenAI guidelines accepted to Nature Human Behaviour, 2025.
• Paper on conditional activation steering accepted to ICLR, 2025 as Spotlight!
• Paper on sufficient explanations accepted to TMLR, 2024.
• Paper (first author) on model agnostic contrastive explanations accepted to IEEE Journal on Emerging and Selected Topics in Circuits and Systems, 2024.
• Paper on constrastive explanations for LLMs (CELL) featured on ABCP news, 2024.
• Paper on causal domain generalization accepted to ECCV, 2024.
• LLM confidence estimation work featured on HuggingFace, 2024.
• Marquis Whos Who Honored Listee, 2024.
• 2 first author papers (one on dynamic sparse training and another on LLM benchmarking) accepted to ACL, 2024 (Findings).
• Gave invited talk and was on panel at Smart Cities and AI Innovation Symposium at UT Austin, 2024.
• First author paper on certifying explanations and identifying trust regions accepted to ICML, 2024.
• Paper on a new dataset for biomedical images accepted to Frontiers in Radiology, 2024.
• Paper on suppressing interpretable concepts accepted to TMLR, 2024.
• First author paper on creating stable and unidirectional explanations accepted to NeurIPS, 2023.
• (Invited) paper in Cell Patterns on diagnosing the current AI ethics debates featured by Montreal AI Ethics Institute, 2023.
• Paper on unsupervised domain adaptation early accepted to MICCAI, 2023.
• AIX360 tutorial with focus on new additions and industrial use cases accepted to KDD, 2023.
• Paper on reprogramming LLMs for antibody infilling accepted to ICML, 2023.
• First author paper on using language to predict odor mixture similarity accepted to Chemical Senses, 2023.
• Paper on practitioner-friendly bias mitigation accepted to FAccT, 2023.
• Paper on clinical toxicity prediction with contrastive explanations accepted to Nature Sci. Reps., 2023.
• Gave invited talk on CoFrNets at Morgan Stanley, 2023.
• Paper studying XAI methods for stance detection accepted to CHIIR, 2023.
• 2 papers (one on single domain generalization and another on RL explainability (first co-author)) accepted to AAAI, 2023.
• CDO magazine awarded our work with an NGO Data4Good award, 2022.
• Extended abstract on collaborative text summarization for chronic pain accepted to ML4H, 2022.
• Paper on contrastive explanations for text accepted to EMNLP, 2022.
• 2 papers on XAI (one first author) accepted to NeurIPS, 2022.
• First author paper on knowledge transfer using a novel multihop approach accepted to IEEE ICKG, 2022.
• Paper on connecting XAI metrics to usage contexts won best paper honorable mention at HCOMP, 2022.
• Gave invited talk on Explainable AI at New England Statistical Society (NESS) Symposium, 2022.
• Gave invited talk on Explainable AI at 2d3d.ai, 2022.
• XAI as applicable to Healthcare and Life Sciences paper accepted to Cell Patterns, 2022.
• Paper on dynamic knowledge transfer accepted to ICLR, 2022.
• Paper on cognitive biases in human decision making accepted to CSCW, 2022.
• Paper on impact (and experiences) of the AIX360 toolkit accepted to IAAI, 2022.