On the Design Space of Discrete Diffusion Online Adaptation for Molecular Optimization

• URL: http://arxiv.org/abs/2607.02834v1
• Date: Fri, 03 Jul 2026 00:06:31 GMT
• Title: On the Design Space of Discrete Diffusion Online Adaptation for Molecular Optimization
• Abstract summary: Molecular optimization often starts from a pretrained generative model that captures a broad prior over valid molecular structures.<n>We study this adaptation problem for discrete diffusion models.<n>We find that acquisition, reward shaping, and model debiasing provide complementary routes to higher reward, especially for small molecules.
• Score: 55.27368653669229
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Molecular optimization often starts from a pretrained generative model that captures a broad prior over valid molecular structures. At test time, however, the goal is not to sample from this prior, but to use a limited oracle budget to shift generation toward task-specific high-reward molecules. We study this adaptation problem for discrete diffusion models. Each online round couples several choices. The loop must decide which candidates to evaluate, how rewards become model updates, which feedback to reuse, and how far to move beyond the pretrained prior. These choices have mostly been studied in isolation, leaving open whether they complement one another, become redundant, or interfere inside a full online adaptation loop. We conduct controlled studies across six small-molecule binding-affinity tasks and three protein-fitness tasks. We find that acquisition, reward shaping, and model debiasing provide complementary routes to higher reward, especially for small molecules. Replay further stabilizes learning, while validity penalties keep small-molecule exploration on the valid molecular manifold. Together, these findings point to a practical recipe for feedback-efficient molecular optimization: online fine-tuning with acquisition, reward shaping, debiasing, replay, and validity control. This recipe outperforms offline fine-tuning and inference-time search baselines under matched oracle-call budgets and GPU-hour accounting. The gains are largest when high-reward candidates require larger shifts from the pretrained prior.

Related papers

• Reference-guided Policy Optimization for Molecular Optimization via LLM Reasoning [58.644854860003704]
  Large language models (LLMs) benefit from supervised fine-tuning (SFT) and reinforcement learning with verifiable rewards (RLVR) in reasoning tasks.<n>We show that answer-only SFT on the reference molecules collapses reasoning, and RLVR provides sparse feedback under similarity constraints.<n>We introduce Reference-guided Policy Optimization (RePO), an optimization approach that learns from reference molecules without requiring trajectory data.
  arXiv  Detail & Related papers  (2026-03-06T04:39:08Z)
• Feedback Efficient Online Fine-Tuning of Diffusion Models [52.170384048274364]
  We propose a novel reinforcement learning procedure that efficiently explores on the manifold of feasible samples. We present a theoretical analysis providing a regret guarantee, as well as empirical validation across three domains.
  arXiv  Detail & Related papers  (2024-02-26T07:24:32Z)
• Augmented Memory: Capitalizing on Experience Replay to Accelerate De Novo Molecular Design [0.0]
  Molecular generative models should learn to satisfy a desired objective under minimal oracle evaluations. We propose a novel algorithm called Augmented Memory that combines data augmentation with experience replay. We show that scores obtained from oracle calls can be reused to update the model multiple times.
  arXiv  Detail & Related papers  (2023-05-10T14:00:50Z)
• Retrieval-based Controllable Molecule Generation [63.44583084888342]
  We propose a new retrieval-based framework for controllable molecule generation. We use a small set of molecules to steer the pre-trained generative model towards synthesizing molecules that satisfy the given design criteria. Our approach is agnostic to the choice of generative models and requires no task-specific fine-tuning.
  arXiv  Detail & Related papers  (2022-08-23T17:01:16Z)
• CELLS: Cost-Effective Evolution in Latent Space for Goal-Directed Molecular Generation [23.618366377098614]
  We propose a cost-effective evolution strategy in latent space, which optimize the molecular latent representation vectors. We adopt a pre-trained molecular generative model to map the latent and observation spaces. We conduct extensive experiments on multiple optimization tasks comparing the proposed framework to several advanced techniques.
  arXiv  Detail & Related papers  (2021-11-30T11:02:18Z)
• Reducing the Long Tail Losses in Scientific Emulations with Active Learning [0.0]
  In this work, we leveraged an active learning approach called core-set selection to actively select data, per a pre-defined budget, to be labelled for training. We tested on two case studies in different fields, namely galaxy halo occupation distribution modelling in astrophysics and x-ray emission spectroscopy in plasma physics.
  arXiv  Detail & Related papers  (2021-11-15T09:02:00Z)
• Flow Network based Generative Models for Non-Iterative Diverse Candidate Generation [110.09855163856326]
  This paper is about the problem of learning a policy for generating an object from a sequence of actions. We propose GFlowNet, based on a view of the generative process as a flow network. We prove that any global minimum of the proposed objectives yields a policy which samples from the desired distribution.
  arXiv  Detail & Related papers  (2021-06-08T14:21:10Z)
• MIMOSA: Multi-constraint Molecule Sampling for Molecule Optimization [51.00815310242277]
  generative models and reinforcement learning approaches made initial success, but still face difficulties in simultaneously optimizing multiple drug properties. We propose the MultI-constraint MOlecule SAmpling (MIMOSA) approach, a sampling framework to use input molecule as an initial guess and sample molecules from the target distribution.
  arXiv  Detail & Related papers  (2020-10-05T20:18:42Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.