https://scholars.lib.ntu.edu.tw/handle/123456789/636993
DC 欄位 | 值 | 語言 |
---|---|---|
dc.contributor.author | Liu, Guan Ting | en_US |
dc.contributor.author | Hu, En Pei | en_US |
dc.contributor.author | PU-JEN CHENG | en_US |
dc.contributor.author | HUNG-YI LEE | en_US |
dc.contributor.author | Sun, Shao Hua | en_US |
dc.date.accessioned | 2023-11-08T07:56:45Z | - |
dc.date.available | 2023-11-08T07:56:45Z | - |
dc.date.issued | 2023-01-01 | - |
dc.identifier.uri | https://scholars.lib.ntu.edu.tw/handle/123456789/636993 | - |
dc.description.abstract | Aiming to produce reinforcement learning (RL) policies that are human-interpretable and can generalize better to novel scenarios, Trivedi et al. (2021) present a method (LEAPS) that first learns a program embedding space to continuously parameterize diverse programs from a pre-generated program dataset, and then searches for a task-solving program in the learned program embedding space when given a task. Despite the encouraging results, the program policies that LEAPS can produce are limited by the distribution of the program dataset. Furthermore, during searching, LEAPS evaluates each candidate program solely based on its return, failing to precisely reward correct parts of programs and penalize incorrect parts. To address these issues, we propose to learn a meta-policy that composes a series of programs sampled from the learned program embedding space. By learning to compose programs, our proposed hierarchical programmatic reinforcement learning (HPRL) framework can produce program policies that describe out-of-distributionally complex behaviors and directly assign credits to programs that induce desired behaviors. The experimental results in the Karel domain show that our proposed framework outperforms baselines. The ablation studies confirm the limitations of LEAPS and justify our design choices. | en_US |
dc.relation.ispartof | Proceedings of Machine Learning Research | en_US |
dc.title | Hierarchical Programmatic Reinforcement Learning via Learning to Compose Programs | en_US |
dc.type | conference paper | en_US |
dc.identifier.scopus | 2-s2.0-85174410787 | - |
dc.identifier.url | https://api.elsevier.com/content/abstract/scopus_id/85174410787 | - |
dc.relation.journalvolume | 202 | en_US |
dc.relation.pageend | 22337 | en_US |
item.openairetype | conference paper | - |
item.openairecristype | http://purl.org/coar/resource_type/c_5794 | - |
item.fulltext | no fulltext | - |
item.grantfulltext | none | - |
item.cerifentitytype | Publications | - |
crisitem.author.dept | Networking and Multimedia | - |
crisitem.author.dept | Computer Science and Information Engineering | - |
crisitem.author.orcid | 0000-0001-5892-0385 | - |
crisitem.author.parentorg | College of Electrical Engineering and Computer Science | - |
crisitem.author.parentorg | College of Electrical Engineering and Computer Science | - |
顯示於: | 電機工程學系 |
在 IR 系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。