Research map / Probabilistic and scientific modeling / Molecules and proteins
Oxytrees: Model Trees for Bipartite Learning
Pedro Ilídio, Felipe Kenji Nakano, Alireza Gharahighehi, Robbe D'hondt, Ricardo Cerri, Celine Vens
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 RLS-Kron leaf models to biclustering trees (vs average-label leaves) (ablation) | Oxytrees · 15 bipartite datasets, inductive | AUPRC/AUROC: Removing the leaf model significantly degraded the performance | Improved |
| add interaction matrix reconstruction (YR) with minimum 5 by 5 leaves and mean leaves (ablation) | Oxytrees · 15 bipartite datasets, inductive | AUPRC/AUROC: worst overall strategy | Hurt |
| use RLS-Kron leaf models with larger minimum leaf sizes up to 20 by 20 (ablation) | Oxytrees · 15 bipartite datasets, inductive | relative score: slight score increase up until 20 by 20 leaves | Improved |
| enforce larger minimum leaf sizes without leaf models (ablation) | forest without leaf models · 15 bipartite datasets, inductive | relative score: drops for leaf dimensions larger than 20 by 20 | Hurt |
| replace PBCT/BICTR split search with proxy-matrix split search | biclustering tree training · synthetic matrices | training time: 35.49 ± 0.75 times faster | Improved |
| use batch leaf-assignment procedure instead of per-pair traversal | biclustering tree inference · synthetic random trees | prediction time: 9.98 ± 0.84 times faster | Improved |
| use Oxytrees instead of BICTR | biclustering forest · 15 datasets, PMP 0% | trees to reach 98% of final score: 42% ± 3% | Improved |
| add tree structure on top of RLS-Kron alone | RLS-Kron · 15 datasets, inductive | AUPRC/AUROC | Improved |
| compare Oxytrees with BICTR in predictive performance | Oxytrees · 15 datasets, inductive | AUPRC/AUROC | No clear effect |
| increase positives masking percentage to 25%, 50%, 75% | all models · 15 datasets | AUPRC/AUROC | Hurt |
| apply NRLMF in inductive instead of transductive setting | NRLMF · 15 datasets | AUPRC/AUROC | Hurt |
Closest papers by meaning
- Proxy-Based Approximation of Shapley and Banzhaf Interactions (NeurIPS 2026 · similarity 0.86)
- PRING: Rethinking Protein-Protein Interaction Prediction from Pairs to Graphs (NeurIPS 2025 · similarity 0.83)
- Towards Foundational Models for Molecular Learning on Large-Scale Multi-Task Datasets (ICLR 2024 · similarity 0.83)
- MixMin: Finding Data Mixtures via Convex Minimization (ICML 2025 · similarity 0.83)
- Pruning neural network models for gene regulatory dynamics using data and domain knowledge (NeurIPS 2024 · similarity 0.82)
- Interventional Causal Discovery in a Mixture of DAGs (NeurIPS 2024 · similarity 0.82)
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.
