DeepRead
Cover of Algorithms for Reinforcement Learning

Algorithms for Reinforcement Learning

Csaba Szepesvari · 2010

Machine learning, statistics and probability › Deep learning and AI

#10,075 most recommended

About the book

The book covers algorithms for reinforcement learning, the problem of learning good behaviour from interaction and reward in Markov decision processes. It is a short monograph in the Synthesis Lectures series, focused on the algorithms themselves rather than on the wider field of machine learning. Its central thread is that a small set of ideas, the Bellman equations and dynamic programming, carries over to settings where the agent learns from sampled experience, and that most practical methods build on this link. It surveys how value prediction and control are approached with temporal-difference learning and Q-learning, both with tabular representations and with function approximation, and it treats policy search as an alternative route. Throughout, it weighs the algorithms by their theoretical properties, such as convergence and sample requirements, alongside their practical behaviour.

Written by Claude from what it knows of this book.

What Hacker News says 67% recommend · 0% critical

In the commenters’ own words

What Open Library readers say

No Open Library reader has rated it yet.

  • 1 want to read
  • 0 reading
  • 0 have read

Open Library’s page for the book

What critics say, via Wikipedia

No Wikipedia article about this book yet, so no critics to quote.

Reddit communities — coming later

Details

First published
2010
ISBN
9781681732138
Editions known
2 at Open Library
Open Library
OL27394907W

Explore in the mosaic

Read next: its neighbours on the mosaic