目录

• Beyond DQN/A3C: A Survey in Advanced Reinforcement Learning(2018)
• Modern Deep Reinforcement Learning Algorithms(2019)
• Evolution Strategies as a Scalable Alternative to Reinforcement Learning
• 模拟器相关
• 代码库
• 迁移学习+强化学习

梳理rl的一些新进展

Beyond DQN/A3C: A Survey in Advanced Reinforcement Learning(2018)

参考深度 | 超越DQN和A3C：深度强化学习领域近期新进展概览

原blog：https://towardsdatascience.com/advanced-reinforcement-learning-6d769f529eb3

DQN

\[ Q\left(s_{t}, a_{t} ; \theta\right) \leftarrow Q\left(s_{t}, a_{t} ; \theta\right)+\alpha[\underbrace{\underbrace{(r_{t}+\max _{a} \hat{Q}\left(s_{t+1}, a ; \theta^{\prime}\right))}_{\text { target }}-Q\left(s_{t}, a_{t} ; \theta\right) )}_{\text {TD-error}}] \]

ac

\[ d \theta_{v} \leftarrow d \theta_{v}+\partial{\underbrace{\left(R-V\left(s_{i} ; \theta_{v}\right)\right)}_{\text{advantage}}}^{2} / \partial \theta_{v} \]

Modern Deep Reinforcement Learning Algorithms(2019)

Modern Deep Reinforcement Learning Algorithms

原论文有点大。。打开太慢。。转存一份：https://daiwk.github.io/assets/Modern%20Deep%20Reinforcement%20Learning%20Algorithms.pdf

Evolution Strategies as a Scalable Alternative to Reinforcement Learning

Evolution Strategies as a Scalable Alternative to Reinforcement Learning

blog：https://openai.com/blog/evolution-strategies/

代码：https://github.com/openai/evolution-strategies-starter

模拟器相关

Virtual-Taobao: Virtualizing Real-world Online Retail Environment for Reinforcement Learning

Simulating User Feedback for Reinforcement Learning Based Recommendations

代码库

https://github.com/p-christ/Deep-Reinforcement-Learning-Algorithms-with-PyTorch

参考17种深度强化学习算法用Pytorch实现

迁移学习+强化学习

八千字长文深度解读，迁移学习在强化学习中的应用及最新进展