Timezone: »

 
Poster
Uncertainty-Aware Attention for Reliable Interpretation and Prediction
Jay Heo · Hae Beom Lee · Saehoon Kim · Juho Lee · Kwang Joon Kim · Eunho Yang · Sung Ju Hwang

Wed Dec 05 02:00 PM -- 04:00 PM (PST) @ Room 210 #62
in Wed Poster Session B »

Attention mechanism is effective in both focusing the deep learning models on relevant features and interpreting them. However, attentions may be unreliable since the networks that generate them are often trained in a weakly-supervised manner. To overcome this limitation, we introduce the notion of input-dependent uncertainty to the attention mechanism, such that it generates attention for each feature with varying degrees of noise based on the given input, to learn larger variance on instances it is uncertain about. We learn this Uncertainty-aware Attention (UA) mechanism using variational inference, and validate it on various risk prediction tasks from electronic health records on which our model significantly outperforms existing attention models. The analysis of the learned attentions shows that our model generates attentions that comply with clinicians' interpretation, and provide richer interpretation via learned variance. Further evaluation of both the accuracy of the uncertainty calibration and the prediction performance with "I don't know'' decision show that UA yields networks with high reliability as well.

Keywords: Variational Inference · Time Series Analysis · Recurrent Networks · Attention Models
[ Paper]  [ Poster] 

Author Information

Jay Heo (KAIST)
Hae Beom Lee (KAIST)
Saehoon Kim (AITRICS)
Juho Lee (University of Oxford)
Kwang Joon Kim (Yonsei University College of Medicine)
Eunho Yang (Korea Advanced Institute of Science and Technology; AItrics)
Sung Ju Hwang (KAIST, AItrics)

More from the Same Authors