← Search

Daksh Idnani

2 accepted papers

2023

Don’t forget the nullspace! Nullspace occupancy as a mechanism for out of distribution failure

ICLR 2023poster

Out of distribution (OoD) generalization has received considerable interest in recent years. In this work, we identify a particular failure mode of OoD generalization for discriminative classifiers that is based on test data (from a new domain) lying in the nullspace of features learnt from source d…

Cited by 3SourcePDFScholar
2021

Usable Information and Evolution of Optimal Representations During Training

ICLR 2021poster

We introduce a notion of usable information contained in the representation learned by a deep network, and use it to study how optimal representations for the task emerge during training. We show that the implicit regularization coming from training with Stochastic Gradient Descent with a high learn…

Cited by 13SourcePDFScholar