发表机构
Northwestern University（西北大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
DiMOS提出免训练的推理时多目标搜索框架，利用Doob引导局部重采样和预算高效轨迹搜索，在DNA、蛋白质和RNA任务上实现最高联合成功率，性能达最强基线的1.98倍。
AI 中文摘要
科学设计通常需要同时满足多个目标和约束。预训练的掩码扩散模型为此任务提供了生成基础，但针对这些目标和约束进行微调会产生额外的训练成本，这促使研究者采用冻结模型的推理时引导方法。然而，此类引导面临两个挑战：通过/失败型约束和黑盒奖励模型可能无法提供有效的梯度，而同时满足多个要求可能导致可行区域很小，使得在有限的推理预算内难以找到可行设计。为应对这些挑战，我们提出了DiMOS，一个免训练的多目标科学设计框架。利用候选补全的联合奖励，DiMOS在无需奖励梯度的情况下执行近似Doob引导的局部重采样。为高效分配计算资源，它采用预算高效的轨迹搜索，将计算集中于有前景的延续路径。在六个DNA、蛋白质和RNA任务中，DiMOS在相近的生成时间内取得了最高的联合成功率，在DNA和蛋白质任务上达到最强基线性能的1.98倍，同时保持了较高的序列独特性和自然性。
英文摘要
Scientific design often requires jointly satisfying multiple objectives and constraints. Pretrained masked diffusion models provide a generative foundation for this task, but fine-tuning them to meet these objectives and constraints incurs additional training costs, motivating inference-time guidance with frozen models. However, such guidance faces two challenges: pass-or-fail constraints and black-box reward models may provide no useful gradients, while jointly satisfying multiple requirements can leave a small feasible region, making feasible designs difficult to find within a limited inference budget. To address these challenges, we introduce DiMOS, a training-free framework for multi-objective scientific design. Using joint rewards from candidate completions, DiMOS performs approximate Doob-guided local resampling without requiring reward gradients. To allocate computation efficiently, it uses budget-efficient trajectory search to focus computation on promising continuations. Across six DNA, protein, and RNA tasks, DiMOS attains the highest joint success rate at comparable generation times, up to $1.98\times$ the strongest baseline on DNA and protein, while maintaining high sequence uniqueness and naturalness.
