Improving Extrapolation Behavior in Deep Neural Networks

Rushil Anirudh | 22-ERD-006

Executive Summary

We will develop capabilities to detect, control, and improve extrapolation and strong generalization behavior in deep neural networks. By enabling neural networks to extrapolate better, we will make deep neural networks better suited for a broad range of scientific explorations, transforming mission-critical research in areas such as biosecurity, physical data modeling, and cognitive simulation.

Publications, Presentations, and Patents

Gokhale, Tejas, Rushil Anirudh, Jayaraman J. Thiagarajan, Bhavya Kailkhura, Chitta Baral, and Yezhou Yang. "Improving Diversity with Adversarially Learned Transformations for Domain Generalization." In Press. Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV). 2023.

Anirudh, Rushil, Jayaraman J. Thiagarajan. "Out of Distribution Detection via Neural Network Anchoring." In Press. Asian Conference on Machine Learning (ACML), PMLR 2022.

Thiagarajan, Jayaraman J., Rushil Anirudh, Vivek Narayanaswamy, and Peer-Timo Bremer. "Single Model Uncertainty Estimation via Stochastic Data Centering." In Press. Advances in Neural Information Processing Systems (NeurIPS), 2022.

Anirudh, R., J.J. Thiagarajan. 2022. AMP OOD Detection (Version 0.1) [Computer software].