Presentation | 2001/11/8 Incremental PDFA Learning for Conversational Agents Masayuki OKAMOTO, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | When finite-state machines are used for dialogue models of a dialogue system, the example-based machine learning technology can be used. For such kind of learning technology, learning algorithms which learn cyclic probabilistic finite-state automata with the state merging method are useful. However, these algorithms should learn the whole data again when the number of learning data set increases. The learning cost is large when algorithms need learn the data every time the number of the data set increases as the gradual dialogue model construction. We proposed a learning method which decreases the re-computation cost with caching the merging information, and evaluated the decision of merging and the learned models. From the comparison among the dialogue models learned from 150 dialogues about the tour-guide task, the method which caches only the renewed states reduced the total number of the state-merging decision by 13% or 24%, though there is little effect by the method which caches all states which need re-compute. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Dialogue Model / Probabilistic DFA / State Merging Method / Incremental Learning |
Paper # | AI2001-43 |
Date of Issue |
Conference Information | |
Committee | AI |
---|---|
Conference Date | 2001/11/8(1days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | |
Vice Chair | |
Secretary | |
Assistant |
Paper Information | |
Registration To | Artificial Intelligence and Knowledge-Based Processing (AI) |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Incremental PDFA Learning for Conversational Agents |
Sub Title (in English) | |
Keyword(1) | Dialogue Model |
Keyword(2) | Probabilistic DFA |
Keyword(3) | State Merging Method |
Keyword(4) | Incremental Learning |
1st Author's Name | Masayuki OKAMOTO |
1st Author's Affiliation | Department of Social Informatics, Kyoto University() |
Date | 2001/11/8 |
Paper # | AI2001-43 |
Volume (vol) | vol.101 |
Number (no) | 419 |
Page | pp.pp.- |
#Pages | 8 |
Date of Issue |