Understanding and Improving Recurrent Networks for Human Activity Recognition by Continuous Attention

TL;DR

This paper introduces two attention mechanisms for recurrent neural networks in human activity recognition, enhancing interpretability and performance by focusing on relevant signals and modalities, and incorporating continuity constraints.

Contribution

It proposes temporal and sensor attention models with continuity constraints, improving interpretability and accuracy in human activity recognition tasks.

Findings

Achieved state-of-the-art results on three datasets.

Attention mechanisms align well with human intuition.

Continuity constraints improve robustness and performance.

Abstract

Deep neural networks, including recurrent networks, have been successfully applied to human activity recognition. Unfortunately, the final representation learned by recurrent networks might encode some noise (irrelevant signal components, unimportant sensor modalities, etc.). Besides, it is difficult to interpret the recurrent networks to gain insight into the models' behavior. To address these issues, we propose two attention models for human activity recognition: temporal attention and sensor attention. These two mechanisms adaptively focus on important signals and sensor modalities. To further improve the understandability and mean F1 score, we add continuity constraints, considering that continuous sensor signals are more robust than discrete ones. We evaluate the approaches on three datasets and obtain state-of-the-art results. Furthermore, qualitative analysis shows that the attention learned by the models agree well with human intuition.