Entity · technique

Q-learning

techniqueactiveq-learning-8521fe68·1 events·first seen May 18, 2026

Aliases: Q-learning

Co-occurring entities

Leslie Pack Kaelbling Divide-and-Conquer Value Learning Berkeley AI Research (BAIR)GRPO PPO Aditya (co-lead author)Temporal Difference Learning Floyd-Warshall Algorithm Goal-Conditioned Reinforcement Learning

More like this (12)

Soft Q-Learning Double Q-learning Imitation Learning self-play reinforcement learning meta-learning Goal-Conditioned Reinforcement Learning Reinforcement Learning Constrained Reinforcement Learning Hierarchical Reinforcement Learning reinforcement learning traffic smoothing Temporal Difference Learning Reinforcement Learning from Human Feedback

Recent events (1)

6Berkeley Ai Research (Bair) Blog·May 18, 2026·source ↗

RL without TD Learning: Divide-and-Conquer Value Learning for Long-Horizon Off-Policy RL

A BAIR blog post introduces a divide-and-conquer paradigm for off-policy reinforcement learning that avoids temporal difference (TD) learning's error accumulation problem by reducing Bellman recursions logarithmically rather than linearly. The approach leverages the triangle inequality structure of goal-conditioned RL to define a transitive Bellman update rule, enabling value learning that scales to long-horizon tasks. The authors claim this is the first practical realization of divide-and-conquer value learning at scale in goal-conditioned RL settings, building on an idea traceable to Kaelbling (1993). The post frames this as a third paradigm alongside TD and Monte Carlo methods, addressing a key gap in scalable off-policy RL.

Evaluation and Benchmarking Agent and Tool Ecosystem Leslie Pack Kaelbling Divide-and-Conquer Value Learning Berkeley AI Research (BAIR)+8 more