July 2024 Presented PAGER, a new approach for proactive failure detection in deep networks, at #ICML2024 [paper] [poster]
July 2024 PRIME is accepted for publication at #ECCV2024. PRIME uses priors from LLMs and VLMs to robustly detect failures in vision models!
July 2024 Interested in obtaining an efficient and effective parameterization of deep networks through implicit neural representations? Our recent preprint studies this! [Paper]
June 2024 Check our new preprint on improving compute efficiency of large deep networks through little companions! [Paper]
June 2024 Preprint on using anchoring to improve generalization of vision transformers is out! [Paper] [Thread]
June 2024 Fresh off the press – New paper on the ICECap effort, which is exploring how ML and HPC technologies could broadly accelerate scientific discovery, is published in Physics of Plasmas journal! [Paper]
June 2024 Presenting our paper “Eyes of a Hawk and Ears of a Fox: Part Prototype Network for Generalized Zero-Shot Learning” in Workshop on Representation Learning with Very Limited Images at #CVPR2024 [Paper]
May 2024 Interested in using deep learned surrogates to close the gap between simulations and experiments? Our recent @MLSTjournal paper provides new insights into transformer adaptation and model selection with extremely limited data!! [Paper]
May 2024 G-DUQ: Presented our paper on calibrating graph neural networks using stochastic centering at #ICLR2024 [Paper] [Project Page]
April 2024 Visited Oregon State University and presented a lecture titled “Towards Safe and Actionable AI: Strategies for Robust Adaptation and Proactive Failure Detection” [Slides]
April 2024 Presented a lecture titled “On Estimating Link Prediction Uncertainty using Stochastic Centering” at #ICASSP2024 [Slides]
April 2024 Presented a poster on our recent work “CREPE: Learnable Prompting With CLIP Improves Visual Relationship Prediction” at #ICASSP2024 [Paper] [Project Page] [Codes]
jjthiagarajan.com 