Research map / Adversarial robustness and security / Certified and empirical robustness
ALERT: Adversarial Learning Enhanced Stability-aware Routing Transformer for Adaptive Depression Detection
Liangyi Kang, Wei Hua, Yan Yang, Jie Liu, Dan Ye
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add Reconstructor 2 (adversarial) (ablation) | ALERT · STMHD | F1: 82.27 -> 84.86 | Improved |
| add Reconstructor 1 and 2 (ablation) | ALERT · STMHD | F1: 80.48 -> 84.86 | Improved |
| add stability coefficient (ablation) | ALERT · STMHD | F1: 84.29 -> 84.86 | Improved |
| add routing and reconstructors (ablation) | ALERT · STMHD | F1: 79.23 -> 84.86 | Improved |
| add Reconstructor 2 (ablation) | ALERT · STMHD | false negative rate: 13.13 -> 9.04 | Improved |
| add reconstructors (adversarial module) (ablation) | ALERT · STMHD | false positive rate: 18.12 -> 14.72 | Improved |
| increase days of posts from 7 to 14 | ALERT · STMHD | F1: 78.33 -> 84.86 | Improved |
| increase days of posts from 14 to 21 | ALERT · STMHD | F1: 84.86 -> 86.19 | Improved |
| use ALERT over best baseline DisorBERT | ALERT · STMHD | F1: 80.64 -> 84.86 | Improved |
| use general-purpose GPT-4 zero-shot instead of task-specific models | GPT-4 · STMHD | F1: 68.73 vs 84.86 | Hurt |
Later papers that compared with this method
1 later accepted paper reports ALERT as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- HogVul: Black-box Adversarial Code Generation Framework Against LM-based Vulnerability Detectors (AAAI 2026 · better on 11 of 11 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Depression Detection via Capsule Networks with Contrastive Learning (AAAI 2024 · similarity 0.82)
- EmotionHallucer: Evaluating Emotion Hallucinations in Multimodal Large Language Models (ICLR 2026 · similarity 0.82)
- EmoNet-Face: An Expert-Annotated Benchmark for Synthetic Emotion Recognition (NeurIPS 2025 · similarity 0.82)
- DepMGNN: Matrixial Graph Neural Network for Video-based Automatic Depression Assessment (AAAI 2025 · similarity 0.82)
- Beyond Mimicking Under-Represented Emotions: Deep Data Augmentation with Emotional Subspace Constraints for EEG-Based Emotion Recognition (AAAI 2024 · similarity 0.81)
- Learning to Route: Per-Sample Adaptive Routing for Multimodal Multitask Prediction (NeurIPS 2025 · similarity 0.80)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
