Modeling meta-cognition for efficient learning
‘Meta-cognitive Radial Basis Function Network’ (McRBFN) and ‘Projection Based Learning’ (PBL) is a machine-learning algorithm used to classify a data sample. Its meta-cognitive component selects one learning strategy from sample deletion, neuron growth and parameter update and sample reservation. Th...
| Glavni avtor: | |
|---|---|
| Drugi avtorji: | |
| Format: | Student Research Paper |
| Jezik: | English |
| Izdano: |
2015
|
| Teme: | |
| Online dostop: | https://hdl.handle.net/10356/105728 http://hdl.handle.net/10220/26034 |
| Izvleček: | ‘Meta-cognitive Radial Basis Function Network’ (McRBFN) and ‘Projection Based Learning’ (PBL) is a machine-learning algorithm used to classify a data sample. Its meta-cognitive component selects one learning strategy from sample deletion, neuron growth and parameter update and sample reservation. The cognitive component adjusts the output weight to minimize the error of prediction using PBL algorithm. In this paper, we propose an improvement on the sample addition strategy in order to prevent the corruption of existing knowledge. At last, we evaluate the improved algorithm using three benchmarking classification problems. |
|---|