Diagonal State Spaces are as Effective as Structured State Spaces

TL;DR

This paper demonstrates that diagonal state space models can match the performance of structured state space models on long-range sequence tasks, offering a simpler and more straightforward approach without sacrificing accuracy.

Contribution

The authors show that diagonal state space models perform as well as structured models like S4 on long-range tasks, simplifying the architecture while maintaining effectiveness.

Findings

Diagonal models match S4 performance on Long Range Arena tasks

Diagonal models perform well on speech classification tasks

Simpler implementation without loss of accuracy

Abstract

Modeling long range dependencies in sequential data is a fundamental step towards attaining human-level performance in many modalities such as text, vision, audio and video. While attention-based models are a popular and effective choice in modeling short-range interactions, their performance on tasks requiring long range reasoning has been largely inadequate. In an exciting result, Gu et al. (ICLR 2022) proposed the $\textit{Structured State Space}$ (S4) architecture delivering large gains over state-of-the-art models on several long-range tasks across various modalities. The core proposition of S4 is the parameterization of state matrices via a diagonal plus low rank structure, allowing efficient computation. In this work, we show that one can match the performance of S4 even without the low rank correction and thus assuming the state matrices to be diagonal. Our $\textit{Diagonal State Space}$ (DSS) model matches the performance of S4 on Long Range Arena tasks, speech classification on Speech Commands dataset, while being conceptually simpler and straightforward to implement.