| 摘要 |
<P>PROBLEM TO BE SOLVED: To provide a pattern learning module having scalability and versatility. Ž<P>SOLUTION: Learning modules 10<SB>1</SB>to 10<SB>N</SB>respectively performs update learning to update a plurality of model parameters of a pattern learning model by using input data. A model parameter sharing part 20 causes the learning modules 10<SB>1</SB>to 10<SB>N</SB>to share the model parameters updated by the respective learning modules of the learning modules 10<SB>1</SB>to 10<SB>N</SB>. A pattern classification part 111 acquires the model parameters of the pattern learning modules after the update learning from the learning modules 10<SB>1</SB>to 10<SB>N</SB>, and classifies the plurality of learning modules on the basis of model parameter distances. This invention may be applied to the learning of time series patterns. Ž<P>COPYRIGHT: (C)2010,JPO&INPIT Ž
|
