Title :
Divide and conquer algorithms for constructing neural networks architectures
Author :
Koutroumbas, K. ; Pouliakis, A. ; Kalouptsidis, N.
Author_Institution :
Dept. of Inf., Univ. of Athens, Athens, Greece
Abstract :
In this paper two algorithms for the construction of feedforward neural network architectures are discussed. Their close relation with a special class of decision tree classifiers is commented. Finally, the performance of the algorithms is assessed on a complex cytological pattern classification application.
Keywords :
decision trees; divide and conquer methods; feedforward neural nets; neural net architecture; pattern classification; complex cytological pattern classification application; decision tree classifiers; divide and conquer algorithms; feedforward neural network architectures; neural networks architectures; Cancer; Computer architecture; Decision trees; Feature extraction; Histograms; Neural networks; Signal processing algorithms; Back Propagation; Cytological Diagnosis; Decision trees; Image Analysis; Image Measurements; Neural networks; Stomach Lesions;
Conference_Titel :
Signal Processing Conference (EUSIPCO 1998), 9th European
Conference_Location :
Rhodes
Print_ISBN :
978-960-7620-06-4
