An introduction to neural computing
An introduction to neural computing
| Hi-index | 0.00 |
This work introducest wonew unsupervised learning algorithms based on the WISARD weightless neural classifier model. The first one, the standard AUTOWISARD model, is able to perform fast one-shot learning of unsorted sets of input patterns. The second one is a recursive version of AUTOWISARD which not only keeps the good features of the basic model, agility, plasticity and stability, but also produces hierarc hically structured tree-like classifications. Although the standard AUTOWISARD model exhibits good classification skills when exposed to symbolic patterns (by producting only few classes containing patterns of different meanings), its recursive version produces even less confusing classes. The stability of both learning algorithms is also demonstrated.