Showing 1–1 of 1 results for author: Murahari, V S

On Attention Models for Human Activity Recognition

Authors: Vishvak S Murahari, Thomas Ploetz

Abstract: Most approaches that model time-series data in human activity recognition based on body-worn sensing (HAR) use a fixed size temporal context to represent different activities. This might, however, not be apt for sets of activities with individ- ually varying durations. We introduce attention models into HAR research as a data driven approach for exploring relevant temporal context. Attention model… ▽ More Most approaches that model time-series data in human activity recognition based on body-worn sensing (HAR) use a fixed size temporal context to represent different activities. This might, however, not be apt for sets of activities with individ- ually varying durations. We introduce attention models into HAR research as a data driven approach for exploring relevant temporal context. Attention models learn a set of weights over input data, which we leverage to weight the temporal context being considered to model each sensor reading. We construct attention models for HAR by adding attention layers to a state- of-the-art deep learning HAR model (DeepConvLSTM) and evaluate our approach on benchmark datasets achieving sig- nificant increase in performance. Finally, we visualize the learned weights to better understand what constitutes relevant temporal context. △ Less

Submitted 19 May, 2018; originally announced May 2018.