Research map / Graph learning / Graph self-supervised learning
NeVAE: A Deep Generative Model for Molecular Graphs
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 |
|---|---|---|---|
| Optimize the NeVAE decoder for penalized logP | NeVAE · Penalized logP, best molecule | penalized logP: 6.82 vs 4.284 | Helped |
| Optimize the NeVAE decoder for QED | NeVAE · QED, best molecule | QED: 0.920 vs 0.913 | Helped |
| Use NeVAE rather than JTVAE as the latent space for Bayesian optimization | Bayesian optimization · Penalized logP, best molecule | penalized logP: 3.503 vs 2.826 | Hurt |
| Use NeVAE rather than JTVAE as the latent space for Bayesian optimization | Bayesian optimization · QED, best molecule | QED: 0.848 vs 0.732 | Hurt |
| Add masking during training and test to prevent invalid valence bonds (ablation) | NeVAE · Triangle-free synthetic graphs | validity: 57% -> 100% | Helped |
| Add masking during test (ablation) | NeVAE trained with masking · Triangle-free synthetic graphs | validity: 68% -> 100% | Helped |
| Add valence masks (ablation) | NeVAE · ZINC | validity: 0.590 -> 1.000 | Helped |
| Add valence masks (ablation) | NeVAE · QM9 | validity: 0.682 -> 0.999 | Helped |
| Increase encoder search depth K (ablation) | NeVAE · Barabási-Albert and Kronecker graphs | model performance | Mixed |
| Use masking during training (ablation) | NeVAE · Molecular graphs | training time: +5% | Hurt |
Closest papers by meaning
- Constrained Graph Variational Autoencoders for Molecule Design (NeurIPS 2018 · similarity 0.92)
- Junction Tree Variational Autoencoder for Molecular Graph Generation (ICML 2018 · similarity 0.92)
- Graph Generative Pre-trained Transformer (ICML 2025 · similarity 0.88)
- 3D Denoisers Are Good 2D Teachers: Molecular Pretraining via Denoising and Cross-Modal Distillation (AAAI 2025 · similarity 0.87)
- Lift Your Molecules: Molecular Graph Generation in Latent Euclidean Space (ICLR 2025 · similarity 0.87)
- Bridging Molecular Graphs and Large Language Models (AAAI 2025 · similarity 0.87)
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.