Multi Resolution Analysis (MRA) for Approximate Self-Attention
This work addresses efficiency challenges in deploying Transformers for natural language processing and vision tasks, presenting an incremental improvement over prior approximation strategies.
The paper tackles the problem of approximating the self-attention matrix in Transformers to improve efficiency, showing that a Multi-Resolution Analysis (MRA)-based approach outperforms most existing efficient self-attention methods for both short and long sequences.
Transformers have emerged as a preferred model for many tasks in natural langugage processing and vision. Recent efforts on training and deploying Transformers more efficiently have identified many strategies to approximate the self-attention matrix, a key module in a Transformer architecture. Effective ideas include various prespecified sparsity patterns, low-rank basis expansions and combinations thereof. In this paper, we revisit classical Multiresolution Analysis (MRA) concepts such as Wavelets, whose potential value in this setting remains underexplored thus far. We show that simple approximations based on empirical feedback and design choices informed by modern hardware and implementation challenges, eventually yield a MRA-based approach for self-attention with an excellent performance profile across most criteria of interest. We undertake an extensive set of experiments and demonstrate that this multi-resolution scheme outperforms most efficient self-attention proposals and is favorable for both short and long sequences. Code is available at \url{https://github.com/mlpen/mra-attention}.