Learning to Predict by the Methods of Temporal Differences
Richard S. Sutton published 'Learning to Predict by the Methods of Temporal Differences' in Machine Learning in 1988. The paper introduces a class of incremental prediction-learning procedures that assign credit using differences between temporally successive predictions and studies convergence and optimality in special cases.
Historical classification
This verified record belongs to LINEAiGE's artificial-intelligence history layer beginning in 1956.
Why it matters
A documented reinforcement-learning research milestone establishing temporal-difference methods in the peer-reviewed literature.
People, organizations & technologies
PEOPLE
Richard S. Sutton
TECHNOLOGIES
Temporal-difference learning, Prediction learning, Reinforcement learning
Evidence & verification
STATE
verified
CONFIDENCE
high
LAST VERIFIED
2026-09-12T06:04:36+08:00
The 1988 Machine Learning paper itself and its bibliographic records establish Richard S. Sutton as author, volume 3 pages 9-44, DOI 10.1007/BF00115009, and the paper's stated introduction of incremental prediction-learning procedures based on differences between temporally successive predictions. LINEAiGE preserves year precision only because the evidence used in this admission run does not justify treating any more specific date as the authoritative publication milestone. No organization, geography, ancestry, influence, causality, or relationship is inferred.
Sources
Learning to Predict by the Methods of Temporal Differences
Machine Learning / Kluwer Academic Publishers
Learning to Predict by the Methods of Temporal Differences
ML Anthology
Trace the chronology
Record identity
LINEAiGE IDtemporal-difference-learning-1988