AI Paper Atlas

Research map / Reinforcement learning

Policy gradient and actor-critic: research map

1,570 papers on Policy gradient and actor-critic in Reinforcement learning, from ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR (2016–2026), grouped into 5 clusters and 18 approaches. The busiest year so far is 2026.

What it is

The workhorse algorithms of deep RL: policy gradients, which push up the probability of good actions, and actor-critic methods, which add a learned value function to guide them. The topic also covers off-policy evaluation from logged data and safe RL under constraints.

Terms you will see on this page:

Policy gradient
updating the policy along the gradient of expected reward.
Actor-critic
a policy (actor) trained with feedback from a learned value function (critic).
Off-policy evaluation (OPE)
estimating a policy's value from data collected by another policy.

Start reading with the most cited papers here:

Within Reinforcement learning, its share held steady from 21.9% in 2023–24 to 20.0% in 2025–26 (343 → 463 papers at ICML, NeurIPS, ICLR, CVPR and AAAI, the venues with data for all four years).

2016: 24162017: 47172018: 100182019: 114192020: 141202021: 153212022: 152222023: 175232024: 184242025: 177252026: 30326

Explore Policy gradient and actor-critic in the interactive map

Working on something in this topic? Describe your idea in scime atlas to see which approach it falls under, the closest papers by meaning and how crowded the spot has become.

What changed recently

Approaches whose share of the topic moved most between 2023–24 and 2025–26 (ICML, NeurIPS, ICLR, CVPR and AAAI):

Papers from 2025–2026 that are already cited most:

Approaches and key papers

Value learning · 813 papers
Keywords: value, td, distributional

Approaches in this cluster:

Most cited and most cited since 2024:

Policy gradients · 254 papers
Keywords: policy gradient, gradients, variance

Approaches in this cluster:

Most cited and most cited since 2024:

Actor-critic methods · 202 papers
Keywords: actor critic, policy actor, critic algorithms

Approaches in this cluster:

Most cited and most cited since 2024:

Safe and constrained RL · 166 papers
Keywords: safe, constrained, constraints

Approaches in this cluster:

Most cited and most cited since 2024:

Off-policy evaluation · 135 papers
Keywords: ope, estimators, importance

Approaches in this cluster:

Most cited and most cited since 2024:

Related topics in Reinforcement learning

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy