Attention-Based LSTM for Psychological Stress Detection from Spoken Language Using Distant Supervision

TL;DR

This paper introduces an attention-based bidirectional LSTM model for detecting psychological stress from spoken language transcriptions, leveraging distant supervision from Twitter hashtags to improve robustness and accuracy.

Contribution

It presents a novel combination of attention-based LSTM with distant supervision for stress detection, expanding training data and enhancing model performance.

Findings

Bidirectional LSTM with attention achieved 74.1% accuracy.

Distant supervision fine-tuning improved accuracy by 1.6%.

Attention mechanism effectively identifies informative words.

Abstract

We propose a Long Short-Term Memory (LSTM) with attention mechanism to classify psychological stress from self-conducted interview transcriptions. We apply distant supervision by automatically labeling tweets based on their hashtag content, which complements and expands the size of our corpus. This additional data is used to initialize the model parameters, and which it is fine-tuned using the interview data. This improves the model's robustness, especially by expanding the vocabulary size. The bidirectional LSTM model with attention is found to be the best model in terms of accuracy (74.1%) and f-score (74.3%). Furthermore, we show that distant supervision fine-tuning enhances the model's performance by 1.6% accuracy and 2.1% f-score. The attention mechanism helps the model to select informative words.