Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Distilling and Adapting: A Topology-Aware Framework for Zero-Shot Interaction Prediction in Multiplex Biological Networks
Alana Deng, Sugitha Janarthanan, Yan Sun, Zihao Jing, Pingzhao Hu
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 |
|---|---|---|---|
| use CAZI-MBN instead of best baseline | multiplex interaction prediction · DGIdb (transductive) | AUROC: 0.551 -> 0.715 | Improved |
| use CAZI-MBN instead of best baseline | multiplex interaction prediction · DGIdb (zero-shot) | AUROC: 0.524 -> 0.671 | Improved |
| use CAZI-MBN instead of best baseline | multiplex interaction prediction · PINNACLE (zero-shot) | AUROC: 0.785 -> 0.812 | Improved |
| use CAZI-MBN instead of best baseline | multiplex interaction prediction · MetaConserve (transductive) | AUROC: 0.737 -> 0.752 | No clear effect |
| use CAZI-MBN instead of best baseline | multiplex interaction prediction · TRRUST (transductive) | Hamming score: 0.883 -> 0.791 | Hurt |
| use CAZI-MBN instead of best baseline | multiplex interaction prediction · MetaConserve (transductive) | subset accuracy: 0.718 -> 0.656 | Hurt |
| use CAZI-MBN instead of best baseline | multiplex interaction prediction · five MBNs, transductive and zero-shot | AUROC/AUPRC: 3.1--20.4% improvement | Improved |
| add domain-specific LLM sequence embeddings (ablation) | CAZI-MBN · five MBNs | AUROC: 15--20% AUROC drop when removed | Improved |
| add context-aware enhancement (CAE) module (ablation) | CAZI-MBN · five MBNs | AUROC: 7--10% AUROC drop when removed | Improved |
| add unified graph tokenizer (UGT) and mixture of experts (MoE) (ablation) | CAZI-MBN · five MBNs | AUROC: 5--8% AUROC drop when removed | Improved |
| use per-label calibrated thresholds post hoc | CAZI-MBN · datasets where CAZI-MBN lags on HS/SA | Hamming score / subset accuracy | Improved |
Closest papers by meaning
- Dual-Channel Learning Framework for Zero-Shot CircRNA-miRNA Interaction Prediction via State Space Modeling (AAAI 2026 · similarity 0.88)
- Evolutionarily Learning Multi-Aspect Interactions and Influences from Network Structure and Node Content (AAAI 2019 · similarity 0.87)
- Revisiting Link Prediction: a data perspective (ICLR 2024 · similarity 0.84)
- Dual-Channel Learning Framework for Drug-Drug Interaction Prediction via Relation-Aware Heterogeneous Graph Transformer (AAAI 2024 · similarity 0.83)
- KINDLE: Knowledge-Guided Distillation for Prior-Free Gene Regulatory Network Inference (NeurIPS 2025 · similarity 0.83)
- Relation-Aware Equivariant Graph Networks for Epitope-Unknown Antibody Design and Specificity Optimization (AAAI 2025 · similarity 0.83)
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.
