Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Uncertainty-aware Semi-supervised Ensemble Teacher Framework for Multilingual Depression Detection

Mohammad Zia Ur Rehman, Velpuru Navya, Sanskar, Shuja Uddin Qureshi, Nagendra Kumar

TL;DR

Depression detection on social media is challenged by multilingual data scarcity and informal language. The authors propose Semi-SMDNet, a semi-supervised multilingual framework that combines a teacher-student setup, ensemble pseudo-labeling (3T), EMA-based stability, uncertainty-weighted training, incremental pseudo-labeling, and data augmentation on four languages. Empirical results across Arabic, Bangla, English, and Spanish show consistent improvements over strong baselines, with particular gains in low-resource settings, and ablation analyses confirm the contribution of each component. The approach demonstrates scalable, cross-language mental health monitoring under limited annotation budgets.

Abstract

Detecting depression from social media text is still a challenging task. This is due to different language styles, informal expression, and the lack of annotated data in many languages. To tackle these issues, we propose, Semi-SMDNet, a strong Semi-Supervised Multilingual Depression detection Network. It combines teacher-student pseudo-labelling, ensemble learning, and augmentation of data. Our framework uses a group of teacher models. Their predictions come together through soft voting. An uncertainty-based threshold filters out low-confidence pseudo-labels to reduce noise and improve learning stability. We also use a confidence-weighted training method that focuses on reliable pseudo-labelled samples. This greatly boosts robustness across languages. Tests on Arabic, Bangla, English, and Spanish datasets show that our approach consistently beats strong baselines. It significantly reduces the performance gap between settings that have plenty of resources and those that do not. Detailed experiments and studies confirm that our framework is effective and can be used in various situations. This shows that it is suitable for scalable, cross-language mental health monitoring where labelled resources are limited.

Uncertainty-aware Semi-supervised Ensemble Teacher Framework for Multilingual Depression Detection

TL;DR

Depression detection on social media is challenged by multilingual data scarcity and informal language. The authors propose Semi-SMDNet, a semi-supervised multilingual framework that combines a teacher-student setup, ensemble pseudo-labeling (3T), EMA-based stability, uncertainty-weighted training, incremental pseudo-labeling, and data augmentation on four languages. Empirical results across Arabic, Bangla, English, and Spanish show consistent improvements over strong baselines, with particular gains in low-resource settings, and ablation analyses confirm the contribution of each component. The approach demonstrates scalable, cross-language mental health monitoring under limited annotation budgets.

Abstract

Detecting depression from social media text is still a challenging task. This is due to different language styles, informal expression, and the lack of annotated data in many languages. To tackle these issues, we propose, Semi-SMDNet, a strong Semi-Supervised Multilingual Depression detection Network. It combines teacher-student pseudo-labelling, ensemble learning, and augmentation of data. Our framework uses a group of teacher models. Their predictions come together through soft voting. An uncertainty-based threshold filters out low-confidence pseudo-labels to reduce noise and improve learning stability. We also use a confidence-weighted training method that focuses on reliable pseudo-labelled samples. This greatly boosts robustness across languages. Tests on Arabic, Bangla, English, and Spanish datasets show that our approach consistently beats strong baselines. It significantly reduces the performance gap between settings that have plenty of resources and those that do not. Detailed experiments and studies confirm that our framework is effective and can be used in various situations. This shows that it is suitable for scalable, cross-language mental health monitoring where labelled resources are limited.
Paper Structure (35 sections, 7 equations, 5 figures, 6 tables)

This paper contains 35 sections, 7 equations, 5 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Key Contributions
  3. Related Work
  4. Methodology
  5. Multilingual Dataset Preparation
  6. Dataset Overview
  7. Data Preprocessing Steps
  8. Data Splits (Labeled, Unlabeled, Test)
  9. Semi-Supervised Learning Framework
  10. Overview of the Student-Teacher Model
  11. Ensemble of Teachers (3T)
  12. Exponential Moving Average (EMA)
  13. Loss Functions and Training Approach
  14. Supervised Loss Calculation
  15. Consistency Loss Calculation
  16. ...and 20 more sections

Figures (5)

  • Figure 1: System Architecture
  • Figure 2: Ablation on data augmentation. Acc_Org and F1_Org represent the accuracy and F1-score of the final model. Acc and F1 represent the values after removing the component.
  • Figure 3: Ablation on IPL. Acc_Org and F1_Org represent the accuracy and F1-score of the final model. Acc and F1 represent the values after removing the component.
  • Figure 4: Ablation on uncertainty weighting. Acc_Org and F1_Org represent the accuracy and F1-score of the final model. Acc and F1 represent the values after removing the component.
  • Figure 5: Ablation on ensemble learning. Acc_Org and F1_Org represent the accuracy and F1-score of the final model. Acc and F1 represent the values after removing the component.