AiR: Attention with Reasoning Capability
This work addresses the need for more interpretable and effective attention mechanisms in AI models, though it is incremental as it builds on existing attention concepts.
The authors tackled the problem of evaluating and improving attention mechanisms in neural networks by proposing the AiR framework, which introduces a metric based on atomic reasoning operations and a supervision method to jointly optimize attention, reasoning, and task performance, resulting in better reasoning capability and task performance.
While attention has been an increasingly popular component in deep neural networks to both interpret and boost performance of models, little work has examined how attention progresses to accomplish a task and whether it is reasonable. In this work, we propose an Attention with Reasoning capability (AiR) framework that uses attention to understand and improve the process leading to task outcomes. We first define an evaluation metric based on a sequence of atomic reasoning operations, enabling quantitative measurement of attention that considers the reasoning process. We then collect human eye-tracking and answer correctness data, and analyze various machine and human attentions on their reasoning capability and how they impact task performance. Furthermore, we propose a supervision method to jointly and progressively optimize attention, reasoning, and task performance so that models learn to look at regions of interests by following a reasoning process. We demonstrate the effectiveness of the proposed framework in analyzing and modeling attention with better reasoning capability and task performance. The code and data are available at https://github.com/szzexpoi/AiR