Hierarchical Multi-scale Attention Networks for action recognition

Yan, Shiyang, Smith, Jeremy S ORCID: 0000-0002-0212-2365, Lu, Wenjin and Zhang, Bailing (2018) Hierarchical Multi-scale Attention Networks for action recognition. SIGNAL PROCESSING-IMAGE COMMUNICATION, 61. pp. 73-84.

Text paper.pdf - Author Accepted Manuscript Download (3MB)

Abstract

Recurrent Neural Networks (RNNs) have been widely used in natural language processing and computer vision. Amongst them, the Hierarchical Multi-scale RNN (HM-RNN), a recently proposed multi-scale hierarchical RNN, can automatically learn the hierarchical temporal structure from data. In this paper, we extend the work to solve the computer vision task of action recognition. However, in sequence-to-sequence models like RNN, it is normally very hard to discover the relationships between inputs and outputs given static inputs. As a solution, the attention mechanism can be applied to extract the relevant information from the inputs thus facilitating the modeling of the input–output relationships. Based on these considerations, we propose a novel attention network, namely Hierarchical Multi-scale Attention Network (HM-AN), by incorporating the attention mechanism into the HM-RNN and applying it to action recognition. A newly proposed gradient estimation method for stochastic neurons, namely Gumbel-softmax, is exploited to implement the temporal boundary detectors and the stochastic hard attention mechanism. To reduce the negative effect of the temperature sensitivity of the Gumbel-softmax, an adaptive temperature training method is applied to improve the system performance. The experimental results demonstrate the improved effect of HM-AN over LSTM with attention on the vision task. Through visualization of what has been learnt by the network, it can be observed that both the attention regions of the images and the hierarchical temporal structure can be captured by a HM-AN.

Item Type:	Article
Uncontrolled Keywords:	Action recognition, Hierarchical multi-scale RNNs, Attention mechanism, Stochastic neurons
Depositing User:	Symplectic Admin
Date Deposited:	30 Nov 2017 10:16
Last Modified:	19 Jan 2023 06:49
DOI:	10.1016/j.image.2017.11.005
Related URLs:	Author Publisher
URI:	https://livrepository.liverpool.ac.uk/id/eprint/3013192