Generalized Invariant Matching Property via LASSO

TL;DR

This paper extends the invariant matching property to settings with target interventions, proposing a Lasso-based algorithm that enhances robustness and computational efficiency in distribution shift scenarios.

Contribution

It generalizes the invariant matching property for target interventions and introduces a Lasso-based method for improved robustness and efficiency.

Findings

The proposed method outperforms existing algorithms in robustness.

It is computationally more efficient due to the Lasso variant.

The approach effectively handles distribution shifts with target interventions.

Abstract

Learning under distribution shifts is a challenging task. One principled approach is to exploit the invariance principle via the structural causal models. However, the invariance principle is violated when the response is intervened, making it a difficult setting. In a recent work, the invariant matching property has been developed to shed light on this scenario and shows promising performance. In this work, by formulating a high-dimensional problem with intrinsic sparsity, we generalize the invariant matching property for an important setting when only the target is intervened. We propose a more robust and computation-efficient algorithm by leveraging a variant of Lasso, improving upon the existing algorithms.