DeepAI AI Chat
Log In Sign Up

Iterative Feature Matching: Toward Provable Domain Generalization with Logarithmic Environments

06/18/2021
by   Yining Chen, et al.
0

Domain generalization aims at performing well on unseen test environments with data from a limited number of training environments. Despite a proliferation of proposal algorithms for this task, assessing their performance, both theoretically and empirically is still very challenging. Moreover, recent approaches such as Invariant Risk Minimization (IRM) require a prohibitively large number of training environments - linear in the dimension of the spurious feature space d_s - even on simple data models like the one proposed by [Rosenfeld et al., 2021]. Under a variant of this model, we show that both ERM and IRM cannot generalize with o(d_s) environments. We then present a new algorithm based on performing iterative feature matching that is guaranteed with high probability to yield a predictor that generalizes after seeing only O(logd_s) environments.

READ FULL TEXT

page 1

page 2

page 3

page 4

01/30/2022

Provable Domain Generalization via Invariant-Feature Subspace Recovery

Domain generalization asks for models trained on a set of training envir...
04/10/2020

An Empirical Study of Invariant Risk Minimization

Invariant risk minimization (IRM; Arjovsky et al., 2019) is a recently p...
02/11/2020

Invariant Risk Minimization Games

The standard risk minimization paradigm of machine learning is brittle w...
01/04/2021

Does Invariant Risk Minimization Capture Invariance?

We show that the Invariant Risk Minimization (IRM) formulation of Arjovs...
06/06/2020

Domain Extrapolation via Regret Minimization

Many real prediction tasks such as molecular property prediction require...
03/23/2023

Rethinking Domain Generalization for Face Anti-spoofing: Separability and Alignment

This work studies the generalization issue of face anti-spoofing (FAS) m...
02/01/2022

Finding lost DG: Explaining domain generalization via model complexity

The domain generalization (DG) problem setting challenges a model traine...