// ♞ Reinforcement Learning

Policy Gradient (REINFORCE)

Directly increases the probability of actions that led to high return.

策略梯度 REINFORCE 1992 Intermediate core

IN PLAIN WORDS

Update by ∇log-prob of action × return; handles continuous and stochastic policies. High variance is reduced with a baseline. It underlies PPO and GRPO.

Used in: Foundation of PPO and GRPO

TIMELINE · prerequisites → this → what builds on it

1951 Stochastic Gr…1992 Policy Gradient (…2015 Trust Region …2015 DDPG / TD3 (C…2016 AlphaGo / Alp…2016 Actor-Critic …2017 Proximal Poli…

01Original paper

Simple statistical gradient-following algorithms for connectionist reinforcement learning

Williams et al. · Machine Learning · 1992

DOI ↗ ✓ Verified · Crossref API doi:10.1007/BF00992696

02Where it sits

Learn the left column first; the right column builds on this one.

Learn first

Stochastic Gradient Descent (SGD)

Policy Gradient (REINFORCE)

Then you can learn

Left: prerequisites · right: builds on this · dashed: related. Full map →

03Learning resources

Official docs, explainers, courses and reference code. All links checked.

Explainers

Lilian Weng：Policy Gradient Algorithms ↗EN

Courses

04Where the course teaches it

U33 Reinforcement Learning Basics MDPs, Q-learning, DQN, policy gradients, PPO and AlphaZero — the prerequisites for post-training and reasoning models. 14 h12 algorithms4 papers

On learning paths:Post-training & Reasoning Models

For AI / data

Verification: verified, checked 2026-10-21.
JSON: /api/learn/algorithms.json (id: policy-gradient)