Self-Attention Neural Bag-of-Features
This work addresses sequence data analysis tasks, but appears incremental as it builds on existing 2D-Attention methodology.
The authors tackled the problem of multivariate sequence data analysis by proposing new self-attention formulations that quantify feature/temporal relevance through latent spaces and a joint 2D attention mechanism, resulting in improved performance compared to standard methods.
In this work, we propose several attention formulations for multivariate sequence data. We build on top of the recently introduced 2D-Attention and reformulate the attention learning methodology by quantifying the relevance of feature/temporal dimensions through latent spaces based on self-attention rather than learning them directly. In addition, we propose a joint feature-temporal attention mechanism that learns a joint 2D attention mask highlighting relevant information without treating feature and temporal representations independently. The proposed approaches can be used in various architectures and we specifically evaluate their application together with Neural Bag of Features feature extraction module. Experiments on several sequence data analysis tasks show the improved performance yielded by our approach compared to standard methods.