Sample Efficient Generative Molecular Optimization with Joint Self-Improvement
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910019329458176 |
|---|---|
| author | Korkmaz, Serra Izdebski, Adam Pirnay, Jonathan Møller-Larsen, Rasmus Kmicikiewicz, Michal Gawade, Pankhil Grimm, Dominik G. Szczurek, Ewa |
| author_facet | Korkmaz, Serra Izdebski, Adam Pirnay, Jonathan Møller-Larsen, Rasmus Kmicikiewicz, Michal Gawade, Pankhil Grimm, Dominik G. Szczurek, Ewa |
| contents | Generative molecular optimization aims to design molecules with properties surpassing those of existing compounds. However, such candidates are rare and expensive to evaluate, yielding sample efficiency essential. Additionally, surrogate models introduced to predict molecule evaluations, suffer from distribution shift as optimization drives candidates increasingly out-of-distribution. To address these challenges, we introduce Joint Self-Improvement, which benefits from (i) a joint generative-predictive model and (ii) a self-improving sampling scheme. The former aligns the generator with the surrogate, alleviating distribution shift, while the latter biases the generative part of the joint model using the predictive one to efficiently generate optimized molecules at inference-time. Experiments across offline and online molecular optimization benchmarks demonstrate that Joint Self-Improvement outperforms state-of-the-art methods under limited evaluation budgets. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_10984 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Sample Efficient Generative Molecular Optimization with Joint Self-Improvement Korkmaz, Serra Izdebski, Adam Pirnay, Jonathan Møller-Larsen, Rasmus Kmicikiewicz, Michal Gawade, Pankhil Grimm, Dominik G. Szczurek, Ewa Machine Learning 68T01 I.2.1 Generative molecular optimization aims to design molecules with properties surpassing those of existing compounds. However, such candidates are rare and expensive to evaluate, yielding sample efficiency essential. Additionally, surrogate models introduced to predict molecule evaluations, suffer from distribution shift as optimization drives candidates increasingly out-of-distribution. To address these challenges, we introduce Joint Self-Improvement, which benefits from (i) a joint generative-predictive model and (ii) a self-improving sampling scheme. The former aligns the generator with the surrogate, alleviating distribution shift, while the latter biases the generative part of the joint model using the predictive one to efficiently generate optimized molecules at inference-time. Experiments across offline and online molecular optimization benchmarks demonstrate that Joint Self-Improvement outperforms state-of-the-art methods under limited evaluation budgets. |
| title | Sample Efficient Generative Molecular Optimization with Joint Self-Improvement |
| topic | Machine Learning 68T01 I.2.1 |
| url | https://arxiv.org/abs/2602.10984 |