Research map / Probabilistic and scientific modeling / Bayesian optimization
Target-Aware Bandit Allocation for Scalable Surrogate Optimization in Chemical Space
Mohammad Haddadnia, Yuvan Chali, Abhilash Jayaraj, Constance Kraay, Joana Reis, Felix Strieth-Kalthoff, Haribabu Arthanari
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 UCB1 instead of epsilon-greedy or softmax to allocate across partitions (ablation) | BOBA · Enamine-5M docked against CKB, top-100 retrieval | top-100 molecules recovered | Improved |
| increase number of partitions K (ablation) | BOBA with UCB1 · Enamine-5M, CKB; also Enamine-S-3.9M and Enamine-HTS | top-100 molecules recovered vs inference cost | Mixed |
| use persistent bandit-allocated partitions instead of a random library subset per iteration (ablation) | BOBA · Enamine-5M docked against CKB | top-100 molecules recovered | Improved |
| use K-means clusters instead of random static partitions as bandit arms (ablation) | BOBA with UCB1 · Enamine-5M docked against CKB | top-100 molecules recovered | Improved |
| cluster in T5Chem embedding space instead of physicochemical descriptor space (ablation) | BOBA · Enamine-5M docked against CKB | top-100 molecules recovered | Improved |
| use Laplace approximation instead of MC dropout or SWAG for surrogate uncertainty (ablation) | BOBA surrogate · Enamine-5M docked against CKB | top-100 molecules recovered | Improved |
| scale library size from about 10^5 to 10^8 molecules | BOBA vs full-library BO · ZINC library docked against AmpC | normalized AUC of top-1000 retrieval | Improved |
| apply BOBA to more diverse library (Enamine-S-3.9M) vs Enamine-5M | BOBA · CKB, NEDD4 | top-100 molecules recovered | Hurt |
Closest papers by meaning
- Efficient Evolutionary Search Over Chemical Space with Large Language Models (ICLR 2025 · similarity 0.86)
- On the Design Space of Discrete Diffusion Online Adaptation for Molecular Optimization (NeurIPS 2026 · similarity 0.86)
- Amortized Molecular Optimization via Group Relative Policy Optimization (NeurIPS 2026 · similarity 0.86)
- Uncertainty-Aware Multi-Objective Reinforcement Learning-Guided Diffusion Models for 3D De Novo Molecular Design (NeurIPS 2025 · similarity 0.86)
- Approximation-Aware Bayesian Optimization (NeurIPS 2024 · similarity 0.85)
- Multi-Objective Molecular Design Through Learning Latent Pareto Set (AAAI 2025 · similarity 0.85)
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.
