[2010.06396] Interpreting Attention Models with Human Visual Attention in Machine Reading Comprehensionopen searchopen navigation menucontact arXivsubscribe to arXiv mailings

While neural networks with attention mechanisms have achieved superior performance on many natural language processing tasks, it remains unclear to which extent learned attention resembles human visual attention. In this paper, we propose a new method that leverages eye-tracking data to investigate the relationship between human visual attention and neural attention in machine reading comprehension. To this end, we introduce a novel 23 participant eye tracking dataset - MQA-RC, in which participants read movie plots and answered pre-defined questions. We compare state of the art networks based on long short-term memory (LSTM), convolutional neural models (CNN) and XLNet Transformer architectures. We find that higher similarity to human attention and performance significantly correlates to the LSTM and CNN models. However, we show this relationship does not hold true for the XLNet models -- despite the fact that the XLNet performs best on this challenging task. Our results suggest that

1 mentions: @shunk031
Keywords: attention
Date: 2020/10/14 05:22

Related Entries

Read more SC-FEGAN:Face Editing Generative Adversarial Networks with Users Sketch and Color - Paper Survey
1 users, 1 mentions 2019/02/21 00:00
Read more Interpretable Adversarial Training for Text - Paper Survey
0 users, 1 mentions 2019/10/28 00:00
Read more [1902.00985] Adversarial Networks and Autoencoders: The Primal-Dual Relationship and Generalization ...
0 users, 1 mentions 2020/02/03 00:51
Read more [1809.02568] Skin lesion classification with ensemble of squeeze-and-excitation networks and semi-su...
0 users, 1 mentions 2020/06/05 05:21
Read more [2006.05656] Why is Attention Not So Attentive?open searchopen navigation menucontact arXivsubscribe...
0 users, 1 mentions 2020/07/05 15:51
Read more 【外部発表】Human Attention Maps for Text Classification: Do Humans and Neural Networks Focus on the Same ...
0 users, 1 mentions 2020/08/19 00:00