样本高效的强化学习与分子设计的积极学习
Michael Dodds1, Jeff Guo1, Thomas Löhr1
1Molecular AI, Discovery Sciences, R&D, AstraZeneca 431 50 Gothenburg Sweden jonpaul.janet@astrazeneca.com.
Chemical science
|March 15, 2024
概括
这项研究引入了一种新的强化学习与主动学习 (RL-AL) 系统,以加速药物发现. RL-AL显著提高了样本效率,加速了在复杂化学空间中寻找新药候选者的速度.
科学领域:
- 计算化学和药物发现
- 科学研究中的人工智能
- 机器学习用于分子设计
背景情况:
- 强化学习 (RL) 对高维问题是有效的,但在复杂的科学环境中,它在样本效率方面存在困难.
- 药物发现需要有效地探索广的化学空间和多参数优化 (MPO).
- 目前的in silico方法,如虚拟选和de novo生成,需要提高复杂模型的样本效率.
研究的目的:
- 为了提高分子设计的强化学习 (RL) 的样本效率.
- 开发和评估一种与RL (RL-AL) 集成的新型主动学习 (AL) 系统,用于多参数优化 (MPO).
- 解决结合RL和AL的挑战,以实现有效的分子发现.
主要方法:
- 积极学习 (AL) 系统与强化学习 (RL) 模型的整合,称为RL-AL.
- 开发一种新的AL方法,专门用于解决分子设计中的多参数优化 (MPO) 问题.
- 与基线RL进行比较分析,使用基于连接体和结构的Oracle函数.
主要成果:
- 对于固定Oracle预算而言,RL-AL显示了生成的访问量增加了5-66倍,计算时间减少了4-64倍.
- 通过RL-AL发现的化合物显示了多参数评分目标的显著丰富,这表明在识别高得分分子方面具有卓越的性能.
- RL-AL方法保持了输出多样性,同时提高了高得分化合物治愈的有效性.
结论:
- 与基线RL相比,RL-AL系统大大加快了寻找新型分子解决方案的速度.
- 这种方法提高了昂贵的计算预言函数的可行性,以前由高成本限制.
- RL-AL方法广泛适用于任何需要提高样本效率和优化的RL领域.
相关概念视频
Structure-Activity Relationships and Drug Design
716
Drug design is a dynamic field that involves discovering and developing new medications based on specific biological targets. This process heavily relies on structure-activity relationships (SAR) and quantitative structure-activity relationships (QSAR) to guide the design and optimization of efficient drugs.
SAR studies the intricate relationship between a drug's chemical structure and biological activity. It focuses on understanding how modifications to a drug's structure can influence...
SAR studies the intricate relationship between a drug's chemical structure and biological activity. It focuses on understanding how modifications to a drug's structure can influence...
716
Mechanistic Models: Compartment Models in Algorithms for Numerical Problem Solving
53
Mechanistic models play a crucial role in algorithms for numerical problem-solving, particularly in nonlinear mixed effects modeling (NMEM). These models aim to minimize specific objective functions by evaluating various parameter estimates, leading to the development of systematic algorithms. In some cases, linearization techniques approximate the model using linear equations.
In individual population analyses, different algorithms are employed, such as Cauchy's method, which uses a...
In individual population analyses, different algorithms are employed, such as Cauchy's method, which uses a...
53


