Name
Abstract | ||
|---|---|---|
It is important to create a set of cooperative learning modules in order to combine them into an ensemble learning machine. If a data point has already been learned by the ensemble learning system, some individual learners in the ensemble should know how to react by either stopping learning further on the learned data, or forgetting the learned data. Such awareness would be essential in keeping each individual learner to be a specialist and cooperative to other individual learners in the same ensemble learning system. In this paper, each individual learning machine gains awareness through adjusting learning directions and learning strength in learning each data by negative correlation learning with selection (NCLS). If the data had not been learned by the ensemble system, each individual learning module in the ensemble would adopt the common learning strategies in which the weight learning would lead the moduleu0027s output to be close to the target output. However, if the data had been learned by the ensemble system, each individual learning module would tend to go to the opposite learning direction, or weaken the learning strength in order to learn to be different to the rest of the learning modules in the ensemble. Simulation results have been shown how such mixed learning strategies could build awareness among the individual learning modules in the ensemble. |
Year | DOI | Venue |
|---|---|---|
2019 | 10.1109/CISP-BMEI48845.2019.8965969 | CISP-BMEI |
DocType | Citations | PageRank |
Conference | 0 | 0.34 |
References | Authors | |
0 | 1 | |