Back
State abstraction in reinforcement learning
Abstract
This work concerns state abstraction - one of commonly proposed solutions to the curse of dimensionality problem. A particular type of state abstraction - state space abstraction is analyzed as a variable selection issue. As an effect of this analysis, an incremental state abstraction algorithm is introduced, inspired by the notions of stimulus discrimination, ambiguity and closure from behavioral psychology. This algorithm correctly solves the variable selection problem by including or removing variables one by one. It is the first among existing solutions to work not only for discrete problems, but also continuous ones.- Record ID
- WUT113e3e486cfe4a4f9971e1d0d5e4db00
- Diploma type
- Doctor of Philosophy
- Author
- Title in Polish
- Abstrakcja stanu w uczeniu ze wzmacnianiem
- Title in English
- State abstraction in reinforcement learning
- Language
- (en) English
- Certifying Unit
- Faculty of Electronics and Information Technology (FEIT)
- Discipline
- automation and robotics / (technology domain) / (technological sciences)
- Status
- Finished
- Defense Date
- 17-11-2015
- Title date
- 24-11-2015
- Supervisor
- Internal reviewers
- External reviewers
- Honored
- yes
- Pages
- 149
- Keywords in English
- state abstraction, Reinforcement Learning
- Abstract in English
- This work concerns state abstraction - one of commonly proposed solutions to the curse of dimensionality problem. A particular type of state abstraction - state space abstraction is analyzed as a variable selection issue. As an effect of this analysis, an incremental state abstraction algorithm is introduced, inspired by the notions of stimulus discrimination, ambiguity and closure from behavioral psychology. This algorithm correctly solves the variable selection problem by including or removing variables one by one. It is the first among existing solutions to work not only for discrete problems, but also continuous ones.
- Thesis file
-
- File: 1
- bpapis thesis.pdf
-
- Uniform Resource Identifier
- https://repo.pw.edu.pl/info/phd/WUT113e3e486cfe4a4f9971e1d0d5e4db00/
- URN
urn:pw-repo:WUT113e3e486cfe4a4f9971e1d0d5e4db00