Enhancing Out-of-Distribution Detection in Natural Language Understanding via Implicit Layer Ensemble

TL;DR

This paper introduces an implicit layer ensemble method using contrastive learning to improve out-of-distribution detection in NLP, leveraging intermediate layer features for better accuracy.

Contribution

It proposes a novel framework that combines intermediate layer representations into a single, more informative feature for OOD detection, surpassing existing methods.

Findings

Significantly improved OOD detection accuracy across multiple datasets

Effective utilization of intermediate layer features through contrastive learning

Outperforms existing single-representation based methods

Abstract

Out-of-distribution (OOD) detection aims to discern outliers from the intended data distribution, which is crucial to maintaining high reliability and a good user experience. Most recent studies in OOD detection utilize the information from a single representation that resides in the penultimate layer to determine whether the input is anomalous or not. Although such a method is straightforward, the potential of diverse information in the intermediate layers is overlooked. In this paper, we propose a novel framework based on contrastive learning that encourages intermediate features to learn layer-specialized representations and assembles them implicitly into a single representation to absorb rich information in the pre-trained language model. Extensive experiments in various intent classification and OOD datasets demonstrate that our approach is significantly more effective than other works.