| 摘要 |
PROBLEM TO BE SOLVED: To generate time-series data with variation. SOLUTION: A score showing a matching degree of each node of an output network net<SB>12</SB>composed of a plurality of nodes having a spatial layout structure to output data is determined, and a node having the highest score is determined as a winner node that is a node most suitable to time-series data. Based on the winner node, the output network net<SB>12</SB>is updated in a self-organized way. Further, based on the time-series data and a spike pattern that is a distribution of the scores to the time series data of the nodes of the output network net<SB>12</SB>, time series-correspondence information for associating the time-series data with the spike pattern is learnt. The date processing method is applicable to, for example, a robot. COPYRIGHT: (C)2008,JPO&INPIT
|
