Equi-Tuning: Group Equivariant Fine-Tuning of Pretrained Models

Authors

  • Sourya Basu University of Illinois at Urbana-Champaign
  • Prasanna Sattigeri IBM Research
  • Karthikeyan Natesan Ramamurthy IBM Research
  • Vijil Chenthamarakshan IBM Research
  • Kush R. Varshney IBM Research
  • Lav R. Varshney University of Illinois at Urbana-Champaign
  • Payel Das IBM Research

DOI:

https://doi.org/10.1609/aaai.v37i6.25832

Keywords:

ML: Deep Neural Architectures, CV: Representation Learning for Vision, ML: Bias and Fairness, SNLP: Bias, Fairness, Transparency & Privacy, SNLP: Generation

Abstract

We introduce equi-tuning, a novel fine-tuning method that transforms (potentially non-equivariant) pretrained models into group equivariant models while incurring minimum L_2 loss between the feature representations of the pretrained and the equivariant models. Large pretrained models can be equi-tuned for different groups to satisfy the needs of various downstream tasks. Equi-tuned models benefit from both group equivariance as an inductive bias and semantic priors from pretrained models. We provide applications of equi-tuning on three different tasks: image classification, compositional generalization in language, and fairness in natural language generation (NLG). We also provide a novel group-theoretic definition for fairness in NLG. The effectiveness of this definition is shown by testing it against a standard empirical method of fairness in NLG. We provide experimental results for equi-tuning using a variety of pretrained models: Alexnet, Resnet, VGG, and Densenet for image classification; RNNs, GRUs, and LSTMs for compositional generalization; and GPT2 for fairness in NLG. We test these models on benchmark datasets across all considered tasks to show the generality and effectiveness of the proposed method.

Downloads

Published

2023-06-26

How to Cite

Basu, S., Sattigeri, P., Natesan Ramamurthy, K., Chenthamarakshan, V., Varshney, K. R., Varshney, L. R., & Das, P. (2023). Equi-Tuning: Group Equivariant Fine-Tuning of Pretrained Models. Proceedings of the AAAI Conference on Artificial Intelligence, 37(6), 6788-6796. https://doi.org/10.1609/aaai.v37i6.25832

Issue

Section

AAAI Technical Track on Machine Learning I