RT Journal Article
T1 C-Mantec: A novel constructive neural network algorithm incorporating competition between neurons.
A1 Subirats Contreras, José Luis
A1 Franco, Leónardo
A1 Jerez-Aragonés, José Manuel
K1 Redes neuronales (Informática)
K1 Algoritmos computacionales
AB C-Mantec is a novel neural network constructive algorithm that combines competition between neuronswith a stable modified perceptron learning rule. The neuron learning is governed by the thermalperceptron rule that ensures stability of the acquired knowledge while the architecture grows and whilethe neurons compete for new incoming information. Competition makes it possible that even after newunits have been added to the network, existing neurons still can learn if the incoming information issimilar to their stored knowledge, and this constitutes a major difference with existing constructingalgorithms. The new algorithm is tested on two different sets of benchmark problems: a Boolean functionset used in logic circuit design and a well studied set of real world problems. Both sets were used to analyzethe size of the constructed architectures and the generalization ability obtained and to compare the resultswith those from other standard and well known classification algorithms. The problem of overfitting isalso analyzed, and a new built-in method to avoid its effects is devised and successfully applied within anactive learning paradigm that filter noisy examples. The results show that the new algorithm generatesvery compact neural architectures with state-of-the-art generalization capabilities.
PB Elsevier
YR 2011
FD 2011-10-18
LK https://hdl.handle.net/10630/30080
UL https://hdl.handle.net/10630/30080
LA eng
NO Subirats, J. L., Franco, L., & Jerez, J. M. (2012). C-Mantec: A novel constructive neural network algorithm incorporating competition between neurons. Neural Networks, 26, 130–140.
NO The authors acknowledge the support from MICIIN (Spain) through grants TIN2008-04985 and TIN2010-16556 (including FEDER funds) and from Junta de Andalucía through grant P08-TIC-04026.
DS RIUMA. Repositorio Institucional de la Universidad de Málaga
RD 19 ene 2026