Environment-biased Feature Ranking for Novelty Detection Robustness

TL;DR

This paper introduces a method for robust novelty detection that identifies and removes environment-biased features to improve detection accuracy across different environmental conditions.

Contribution

It proposes a feature ranking technique based on environment variance to enhance novelty detection robustness, addressing spurious correlations.

Findings

Improves detection performance by up to 6%

Effectively removes environment-biased features

Validates on real and synthetic benchmarks

Abstract

We tackle the problem of robust novelty detection, where we aim to detect novelties in terms of semantic content while being invariant to changes in other, irrelevant factors. Specifically, we operate in a setup with multiple environments, where we determine the set of features that are associated more with the environments, rather than to the content relevant for the task. Thus, we propose a method that starts with a pretrained embedding and a multi-env setup and manages to rank the features based on their environment-focus. First, we compute a per-feature score based on the feature distribution variance between envs. Next, we show that by dropping the highly scored ones, we manage to remove spurious correlations and improve the overall performance by up to 6%, both in covariance and sub-population shift cases, both for a real and a synthetic benchmark, that we introduce for this task.