Bayesian Attention Modules
This addresses a bottleneck in attention-based models for researchers and practitioners by enabling stochastic attention without major changes, though it is incremental as it builds on existing attention frameworks.
The paper tackled the optimization issues of stochastic attention modules by proposing a scalable, easy-to-implement Bayesian version with simplex-constrained distributions, resulting in consistent improvements across multiple tasks like graph node classification and machine translation.
Attention modules, as simple and effective tools, have not only enabled deep neural networks to achieve state-of-the-art results in many domains, but also enhanced their interpretability. Most current models use deterministic attention modules due to their simplicity and ease of optimization. Stochastic counterparts, on the other hand, are less popular despite their potential benefits. The main reason is that stochastic attention often introduces optimization issues or requires significant model changes. In this paper, we propose a scalable stochastic version of attention that is easy to implement and optimize. We construct simplex-constrained attention distributions by normalizing reparameterizable distributions, making the training process differentiable. We learn their parameters in a Bayesian framework where a data-dependent prior is introduced for regularization. We apply the proposed stochastic attention modules to various attention-based models, with applications to graph node classification, visual question answering, image captioning, machine translation, and language understanding. Our experiments show the proposed method brings consistent improvements over the corresponding baselines.