Out-of-Distribution Detection with Attention Head Masking for Multimodal Document Classification

TL;DR

This paper introduces attention head masking (AHM), a novel method for out-of-distribution detection in multi-modal document classification, which outperforms existing approaches and significantly reduces false positives.

Contribution

The paper proposes AHM for multi-modal OOD detection in documents and introduces the FinanceDocs dataset to facilitate further research.

Findings

AHM outperforms state-of-the-art OOD detection methods

AHM reduces false positive rate by up to 7.5%

FinanceDocs dataset supports future research in document OOD detection

Abstract

Detecting out-of-distribution (OOD) data is crucial in machine learning applications to mitigate the risk of model overconfidence, thereby enhancing the reliability and safety of deployed systems. The majority of existing OOD detection methods predominantly address uni-modal inputs, such as images or texts. In the context of multi-modal documents, there is a notable lack of extensive research on the performance of these methods, which have primarily been developed with a focus on computer vision tasks. We propose a novel methodology termed as attention head masking (AHM) for multi-modal OOD tasks in document classification systems. Our empirical results demonstrate that the proposed AHM method outperforms all state-of-the-art approaches and significantly decreases the false positive rate (FPR) compared to existing solutions up to 7.5\%. This methodology generalizes well to multi-modal data, such as documents, where visual and textual information are modeled under the same Transformer architecture. To address the scarcity of high-quality publicly available document datasets and encourage further research on OOD detection for documents, we introduce FinanceDocs, a new document AI dataset. Our code and dataset are publicly available.