Research map / Probabilistic and scientific modeling / Molecules and proteins
Discriminative Embeddings of Latent Variable Models for Structured Data
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 |
|---|---|---|---|
| Use DE-MF/DE-LBP embeddings instead of best string kernel | structure2vec · SCOP | mean AUC: 0.8662 -> 0.9167 | Helped |
| Use structure2vec instead of string kernels on RNA dataset | structure2vec · FC RES | mean AUC: 0.7690 -> 0.7713 | Helped |
| Concatenate k-mer kernels instead of single k (ablation) | spectrum string kernel · SCOP | mean AUC: 0.7097 -> 0.8467 | Helped |
| Concatenate k-mer kernels instead of single k (ablation) | spectrum string kernel · FC RES | mean AUC: 0.7606 -> 0.7576 | No clear effect |
| Replace WL lv-6 with DE-LBP | CEP regression · CEP | test MAE: 0.0962 -> 0.0850 | Helped |
| Replace WL lv-3 with DE-LBP | CEP regression · CEP | test MAE: 0.1431 -> 0.0850 | Helped |
| Replace mean-field with loopy BP embedding (ablation) | structure2vec · CEP | test RMSE: 0.1250 -> 0.1174 | Helped |
| Use structure2vec instead of WL lv-6 in model size | CEP · CEP | # params: 1378m -> 0.1m | Helped |
| Use graph embedding instead of mean predictor | CEP · CEP | test MAE: 1.9864 -> 0.0850 | Helped |
| Increase fixed point iterations (ablation) | structure2vec · CEP | convergence speed | Helped |
| Run only one fixed point iteration (ablation) | DE-MF · CEP | test error | Hurt |
| Use structure2vec on graph kernel benchmarks | structure2vec · MUTAG, NCI1, NCI109, ENZYMES | accuracy | No clear effect |
Closest papers by meaning
- Learning Structured Latent Factors from Dependent Data:A Generative Model Framework from Information-Theoretic Perspective (ICML 2020 · similarity 0.85)
- Optimal Representation Size: High-Dimensional Analysis of Pretraining and Linear Probing (NeurIPS 2026 · similarity 0.85)
- Identifying Metric Structures of Deep Latent Variable Models (ICML 2025 · similarity 0.84)
- Community Detection Guarantees using Embeddings Learned by Node2Vec (NeurIPS 2024 · similarity 0.84)
- Domain2Vec: Vectorizing Datasets to Find the Optimal Data Mixture without Training (ICML 2025 · similarity 0.84)
- Structural Entropy Guided Probabilistic Coding (AAAI 2025 · similarity 0.84)
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.