Research map / Label-efficient and robust learning / Recognition and long-tail
Learning Invariant Molecular Representation in Latent Discrete Space
Xiang Zhuang, Qiang Zhang, Keyan Ding, Yatao Bian, Xiao Wang, Jingsong Lv, Hongyang Chen, Huajun Chen
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 RVQ module to ERM (ablation) | ERM · GOOD-HIV scaffold covariate | ROC-AUC: 69.55 -> 70.18 | Improved |
| Add RVQ module to ERM (ablation) | ERM · GOOD-PCBA size concept | AP: 15.60 -> 16.59 | Improved |
| Remove VQ module (ablation) | Ours (RVQ invariant learning) · GOOD-HIV scaffold covariate | ROC-AUC: 72.93 -> 70.18 | Hurt |
| Remove VQ module (ablation) | Ours (RVQ invariant learning) · GOOD-PCBA size concept | AP: 18.21 -> 17.37 | Hurt |
| Remove residual connection (ablation) | Ours (RVQ invariant learning) · GOOD-PCBA size concept | AP: 18.21 -> 12.99 | Hurt |
| Remove residual connection (ablation) | Ours (RVQ invariant learning) · GOOD-HIV scaffold covariate | ROC-AUC: 72.93 -> 72.40 | Hurt |
| Remove invariant learning objective L_inv (ablation) | Ours (RVQ invariant learning) · GOOD-HIV scaffold covariate | ROC-AUC: 72.93 -> 70.73 | Hurt |
| Remove regularization L_reg (ablation) | Ours (RVQ invariant learning) · GOOD-PCBA size covariate | AP: 18.02 -> 17.33 | Hurt |
| Remove commitment loss L_cmt (ablation) | Ours (RVQ invariant learning) · GOOD-HIV scaffold covariate | ROC-AUC: 72.93 -> 70.20 | Hurt |
| Replace own invariant objective with CIGA objective (ablation) | Ours (RVQ invariant learning) · GOOD-HIV scaffold covariate | ROC-AUC: 72.93 -> 71.23 | Hurt |
| Replace own invariant objective with GREA objective (ablation) | Ours (RVQ invariant learning) · GOOD-PCBA size concept | AP: 18.21 -> 16.86 | Hurt |
| Increase codebook size (ablation) | Ours · GOOD-HIV scaffold covariate | ROC-AUC | Mixed |
Closest papers by meaning
- Enhancing Molecular Representation Learning Through the Combination of 3D and 2D Graph Machine Learning (Student Abstract) (AAAI 2025 · similarity 0.87)
- A probability contrastive learning framework for 3D molecular representation learning (NeurIPS 2024 · similarity 0.86)
- Learning Disentangled Equivariant Representation for Explicitly Controllable 3D Molecule Generation (AAAI 2025 · similarity 0.86)
- Learning Molecular Representation in a Cell (ICLR 2025 · similarity 0.86)
- UniCorn: A Unified Contrastive Learning Approach for Multi-view Molecular Representation Learning (ICML 2024 · similarity 0.86)
- Molecular Data Programming: Towards Molecule Pseudo-labeling with Systematic Weak Supervision (CVPR 2024 · similarity 0.85)
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.
