conftrace_

2021 NeurIPS NeurIPS 2021

Editing a classifier by rewriting its prediction rules

Abstract

We propose a methodology for modifying the behavior of a classifier by directly rewriting its prediction rules. Our method requires virtually no additional data collection and can be applied to a variety of settings, including adapting a model to new environments, and modifying it to ignore spurious features.

Authors

Shibani Santurkar, Dimitris Tsipras, Mahalaxmi Elango, David Bau, Antonio Torralba, Aleksander Madry

Topics

Artificial Intelligence > Core AI > Interpretability Machine Learning > Application Areas > Domain Adaptation

Keywords

domain adaptation spurious feature prediction rule classifier editing

More from NeurIPS 2021

All NeurIPS 2021 papers →

Adaptive Diffusion in Graph Neural Networks

Recurrent Submodular Welfare and Matroid Blocking Semi-Bandits

Representer Point Selection via Local Jacobian Expansion for Post-hoc Classifier Explanation of Deep Neural Networks and Ensemble Models

How Modular should Neural Module Networks Be for Systematic Generalization?

Contrast and Mix: Temporal Contrastive Video Domain Adaptation with Background Mixing

The Flip Side of the Reweighted Coin: Duality of Adaptive Dropout and Regularization