Jethro's Braindump

Neuroscience and Reinforcement Learning

tags
Reinforcement Learning ⭐, Neuroscience ⭐

Reinforcement Learning ⭐ is similar to how humans learn:

  • Learning by trial-and-error
  • Reward is often delayed

TD errors (Temporal Difference Learning) model the activity of dopamine neurons (Schultz, Dayan, and Montague 1997)

Bibliography

Schultz, W., P. Dayan, and P. R. Montague. 1997. “A Neural Substrate of Prediction and Reward.” Science 275 (5306):1593–99.