GENERALIZED DELTA RULE WITH ENTROPY ERROR FUNCTION

Javier BILBAO, Imanol BILBAO, Cristina FENISER

Abstract


Machine Learning can refer to a different and various algorithms, based on artificial intelligence, that are able to recognize data patterns through continuous and repeated learning techniques, and we do not have to assume any prior data distribution. In this way, artificial neural networks can provide an effective technology and methodology. Artificial neural networks need a rule, delta rule, in order to learn from the supplied data. In this paper a generalized delta rule with a new error function is presented.

Keywords: neural networks, delta rule, error function, cost function.


Full Text:

PDF

References


Rumelhart, D.E., McClelland, J.L., Paralell Distributed Processing: Explorations in the Microstructure of Cognition, vol. 1, Cambridge, MA: MIT Press, ISBN 0-262-18120-7, 1986.

Ng, A. CS229 Lecture notes. Machine Learning. Supervised Learning, Discriminative Algorithms http://cs229.stanford.edu/notes/cs229-notes1.pdf


Refbacks

  • There are currently no refbacks.


JOURNAL INDEXED IN :