Reinforcement Learning An Introduction - Richard S. Sutton , Andrew G. Barto.pdf

(3675 KB) Pobierz
Reinforcement
Learning:
An Introduction
Richard S. Sutton
and
Andrew G. Barto
MIT Press,
Cambridge, MA,
1998
A Bradford Book
Endorsements Code Solutions Figures
Errata Course Slides
This introductory textbook on reinforcement learning is targeted toward engineers and
scientists in artificial intelligence, operations research, neural networks, and control
systems, and we hope it will also be of interest to psychologists and neuroscientists.
If you would like to order a copy of the book, or if you are qualified instructor and would
like to see an examination copy, please see the
MIT Press home page for this book.
Or you
might be interested in the reviews at
amazon.com.
There is also a Japanese translation
available.
The table of contents of the book is given below, with associated HTML. The HTML
version has a number of presentation problems, and its text is slightly different from the
real book, but it may be useful for some purposes.
q
Preface
Part I: The Problem
q
1
Introduction
r
1.1 Reinforcement Learning
r
1.2 Examples
r
1.3 Elements of Reinforcement Learning
r
r
r
r
1.4 An Extended Example: Tic-Tac-Toe
1.5 Summary
1.6 History of Reinforcement Learning
1.7 Bibliographical Remarks
q
2
Evaluative Feedback
r
2.1 An n-armed Bandit Problem
r
2.2 Action-Value Methods
r
2.3 Softmax Action Selection
r
2.4 Evaluation versus Instruction
r
2.5 Incremental Implementation
r
2.6 Tracking a Nonstationary Problem
r
2.7 Optimistic Initial Values
r
2.8 Reinforcement Comparison
r
2.9 Pursuit Methods
r
2.10 Associative Search
r
2.11 Conclusion
r
2.12 Bibliographical and Historical Remarks
3
The Reinforcement Learning Problem
r
3.1 The Agent-Environment Interface
r
3.2 Goals and Rewards
r
3.3 Returns
r
3.4 A Unified Notation for Episodic and Continual Tasks
r
3.5 The Markov Property
r
3.6 Markov Decision Processes
r
3.7 Value Functions
r
3.8 Optimal Value Functions
r
3.9 Optimality and Approximation
r
3.10 Summary
r
3.11 Bibliographical and Historical Remarks
Part II: Elementary Methods
q
q
4
Dynamic Programming
r
4.1 Policy Evaluation
r
4.2 Policy Improvement
r
4.3 Policy Iteration
r
4.4 Value Iteration
r
r
r
r
r
4.5 Asynchronous Dynamic Programming
4.6 Generalized Policy Iteration
4.7 Efficiency of Dynamic Programming
4.8 Summary
4.9 Historical and Bibliographical Remarks
q
5
Monte Carlo Methods
r
5.1 Monte Carlo Policy Evaluation
r
5.2 Monte Carlo Estimation of Action Values
r
5.3 Monte Carlo Control
r
5.4 On-Policy Monte Carlo Control
r
5.5 Evaluating One Policy While Following Another
r
5.6 Off-Policy Monte Carlo Control
r
5.7 Incremental Implementation
r
5.8 Summary
r
5.9 Historical and Bibliographical Remarks
6
Temporal Difference Learning
r
6.1 TD Prediction
r
6.2 Advantages of TD Prediction Methods
r
6.3 Optimality of TD(0)
r
6.4 Sarsa: On-Policy TD Control
r
6.5 Q-learning: Off-Policy TD Control
r
6.6 Actor-Critic Methods (*)
r
6.7 R-Learning for Undiscounted Continual Tasks (*)
r
6.8 Games, After States, and other Special Cases
r
6.9 Conclusions
r
6.10 Historical and Bibliographical Remarks
Part III: A Unified View
q
q
7
Eligibility Traces
r
7.1 n-step TD Prediction
r
7.2 The Forward View of TD()
r
7.3 The Backward View of TD()
r
7.4 Equivalence of the Forward and Backward Views
r
7.5 Sarsa()
r
7.6 Q()
r
7.7 Eligibility Traces for Actor-Critic Methods (*)
r
r
r
r
r
7.8 Replacing Traces
7.9 Implementation Issues
7.10 Variable (*)
7.11 Conclusions
7.12 Bibliographical and Historical Remarks
q
8
Generalization and Function Approximation
r
8.1 Value Prediction with Function Approximation
r
8.2 Gradient-Descent Methods
r
8.3 Linear Methods
s
8.3.1 Coarse Coding
s
8.3.2 Tile Coding
s
8.3.3 Radial Basis Functions
s
8.3.4 Kanerva Coding
r
8.4 Control with Function Approximation
r
8.5 Off-Policy Bootstrapping
r
8.6 Should We Bootstrap?
r
8.7 Summary
r
8.8 Bibliographical and Historical Remarks
9
Planning and Learning
r
9.1 Models and Planning
r
9.2 Integrating Planning, Acting, and Learning
r
9.3 When the Model is Wrong
r
9.4 Prioritized Sweeping
r
9.5 Full vs. Sample Backups
r
9.6 Trajectory Sampling
r
9.7 Heuristic Search
r
9.8 Summary
r
9.9 Historical and Bibliographical Remarks
10
Dimensions
r
10.1 The Unified View
r
10.2 Other Frontier Dimensions
11
Case Studies
r
11.1 TD-Gammon
r
11.2 Samuel's Checkers Player
r
11.3 The Acrobot
q
q
q
r
r
r
11.4 Elevator Dispatching
11.5 Dynamic Channel Allocation
11.6 Job-Shop Scheduling
q
q
References
Summary of Notation

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika:

Zgłoś jeśli naruszono regulamin