Research map / Label-efficient and robust learning / Recognition and long-tail
Learning Differences Between Visual Scanning Patterns Can Disambiguate Bipolar and Unipolar Patients
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| replace logistic regression with LSTM RNN on semantic RoI encodings | 8 semantic RoI classifier · BD vs MDD classification (leave-one-out) | AUC: 0.603 -> 0.878 | Helped |
| replace logistic regression with LSTM RNN on grid RoI encodings | 4x4 grid RoI classifier · BD vs MDD classification (leave-one-out) | AUC: 0.450 -> 0.828 | Helped |
| replace logistic regression with LSTM RNN on semantic RoIs | 8 semantic RoI classifier · BD vs MDD classification (leave-one-out) | balanced accuracy: 0.581 -> 0.821 | Helped |
| replace logistic regression with LSTM RNN on grid RoIs | 4x4 grid RoI classifier · BD vs MDD classification (leave-one-out) | balanced accuracy: 0.471 -> 0.773 | Helped |
| encode fixations with saliency maps and CNN (LRCN) instead of manual semantic RoIs | RNN classifier · BD vs MDD classification (leave-one-out) | AUC: 0.878 -> 0.879 | No clear effect |
| encode fixations with LRCN instead of semi-automatic grid RoIs | RNN classifier · BD vs MDD classification (leave-one-out) | AUC: 0.828 -> 0.879 | Helped |
| use 2 instead of 1 convolutional layer in LRCN feature extractor (ablation) | LRCN · BD vs MDD classification (leave-one-out) | AUC: 0.869 -> 0.872 | No clear effect |
| use 3 instead of 2 convolutional layers in LRCN feature extractor (ablation) | LRCN · BD vs MDD classification (leave-one-out) | AUC: 0.872 -> 0.879 | No clear effect |
| use 3 instead of 1 convolutional layer in LRCN feature extractor (ablation) | LRCN · BD vs MDD classification (leave-one-out) | balanced accuracy: 0.771 -> 0.801 | Mixed |
| use 8 instead of 5 semantic RoIs (ablation) | RNN w/ semantic RoIs · BD vs MDD classification (leave-one-out) | AUC: 0.872 -> 0.878 | No clear effect |
| use 8 instead of 5 semantic RoIs (ablation) | RNN w/ semantic RoIs · BD vs MDD classification (leave-one-out) | balanced accuracy: 0.802 -> 0.821 | No clear effect |
| use 4x4 instead of 3x3 grid RoIs (ablation) | RNN w/ grid RoIs · BD vs MDD classification (leave-one-out) | AUC: 0.823 -> 0.828 | No clear effect |
Closest papers by meaning
- Dep-MAP: A Multi-level Alignment Framework with Semantic Prototypes for Video-based Automatic Depression Assessment (AAAI 2026 · similarity 0.84)
- Mental-Perceiver: Audio-Textual Multi-Modal Learning for Estimating Mental Disorders (AAAI 2025 · similarity 0.82)
- NeuroTree: Hierarchical Functional Brain Pathway Decoding for Mental Health Disorders (ICML 2025 · similarity 0.81)
- DepMGNN: Matrixial Graph Neural Network for Video-based Automatic Depression Assessment (AAAI 2025 · similarity 0.81)
- PathVQ: Reforming Computational Pathology Foundation Model for Whole Slide Image Analysis via Vector Quantization (NeurIPS 2025 · similarity 0.81)
- NSD-Imagery: A Benchmark Dataset for Extending fMRI Vision Decoding Methods to Mental Imagery (CVPR 2025 · similarity 0.81)
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.