Action selection methods using reinforcement learning
The Action Selection problem is the problem of run-time choice between conflicting and heterogeneous goals, a central problem in the simulation of whole creatures (as opposed to the solution of isolated uninterrupted tasks). This thesis argues that Reinforcement Learning has been overlooked in the s...
| Main Author: | |
|---|---|
| Published: |
University of Cambridge
1996
|
| Subjects: | |
| Online Access: | http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.604775 |
| id |
ndltd-bl.uk-oai-ethos.bl.uk-604775 |
|---|---|
| record_format |
oai_dc |
| spelling |
ndltd-bl.uk-oai-ethos.bl.uk-6047752017-12-24T16:12:09ZAction selection methods using reinforcement learningHumphrys, Mark1996The Action Selection problem is the problem of run-time choice between conflicting and heterogeneous goals, a central problem in the simulation of whole creatures (as opposed to the solution of isolated uninterrupted tasks). This thesis argues that Reinforcement Learning has been overlooked in the solution of the Action Selection problem. Considering a decentralised model of mind, with internal tension and competition between selfish behaviors, this thesis introduces an algorithm called "W-learning", whereby different parts of the mind modify their behavior based on whether or not they are succeeding in getting the body to execute their actions. This thesis sets W-learning in context among the different ways of exploiting Reinforcement Learning numbers for the purposes of Action Selection. It is a 'Minimize the Worst Unhappiness' strategy. The different methods are tested and their strengths and weaknesses analysed in an artificial world.006.3University of Cambridgehttp://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.604775https://www.repository.cam.ac.uk/handle/1810/252269Electronic Thesis or Dissertation |
| collection |
NDLTD |
| sources |
NDLTD |
| topic |
006.3 |
| spellingShingle |
006.3 Humphrys, Mark Action selection methods using reinforcement learning |
| description |
The Action Selection problem is the problem of run-time choice between conflicting and heterogeneous goals, a central problem in the simulation of whole creatures (as opposed to the solution of isolated uninterrupted tasks). This thesis argues that Reinforcement Learning has been overlooked in the solution of the Action Selection problem. Considering a decentralised model of mind, with internal tension and competition between selfish behaviors, this thesis introduces an algorithm called "W-learning", whereby different parts of the mind modify their behavior based on whether or not they are succeeding in getting the body to execute their actions. This thesis sets W-learning in context among the different ways of exploiting Reinforcement Learning numbers for the purposes of Action Selection. It is a 'Minimize the Worst Unhappiness' strategy. The different methods are tested and their strengths and weaknesses analysed in an artificial world. |
| author |
Humphrys, Mark |
| author_facet |
Humphrys, Mark |
| author_sort |
Humphrys, Mark |
| title |
Action selection methods using reinforcement learning |
| title_short |
Action selection methods using reinforcement learning |
| title_full |
Action selection methods using reinforcement learning |
| title_fullStr |
Action selection methods using reinforcement learning |
| title_full_unstemmed |
Action selection methods using reinforcement learning |
| title_sort |
action selection methods using reinforcement learning |
| publisher |
University of Cambridge |
| publishDate |
1996 |
| url |
http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.604775 |
| work_keys_str_mv |
AT humphrysmark actionselectionmethodsusingreinforcementlearning |
| _version_ |
1718574608238510080 |