Activation Subspaces for Out-of-Distribution Detection

TL;DR

This paper introduces ActSub, a novel OOD detection method that leverages singular value decomposition of model weights to identify decisive and insignificant activation subspaces, improving detection especially under large distribution shifts.

Contribution

The paper proposes a new OOD detection approach using activation subspace decomposition, achieving state-of-the-art results across multiple benchmarks.

Findings

Insignificant subspace better detects Far-OOD data.

Decisive subspace enhances Near-OOD detection.

ActSub outperforms existing methods on standard benchmarks.

Abstract

To ensure the reliability of deep models in real-world applications, out-of-distribution (OOD) detection methods aim to distinguish samples close to the training distribution (in-distribution, ID) from those farther away (OOD). In this work, we propose a novel OOD detection method that utilizes singular value decomposition of the weight matrix of the classification head to decompose the model's activations into decisive and insignificant components, which contribute maximally, respectively minimally, to the final classifier output. We find that the subspace of insignificant components more effectively distinguishes ID from OOD data than raw activations in regimes of large distribution shifts (Far-OOD). This occurs because the classification objective leaves the insignificant subspace largely unaffected, yielding features that are ''untainted'' by the target classification task. Conversely, in regimes of smaller distribution shifts (Near-OOD), we find that activation shaping methods profit from only considering the decisive subspace, as the insignificant component can cause interference in the activation space. By combining two findings into a single approach, termed ActSub, we achieve state-of-the-art results in various standard OOD benchmarks.