Training neural networks with heterogeneous data

• Published in: Neural networks 2005-07, Vol.18 (5), p.595-601
• Main Authors: Drakopoulos, John A., Abdulkader, Ahmad
• Format: Article
• Language: English
• Subjects: Algorithms; Applied sciences; Artificial Intelligence; Boosting; Classification; Computer science; control theory; systems; Connectionism. Neural networks; Data emphasizing; Data Interpretation, Statistical; Exact sciences and technology; Game Theory; Growing cell structure; Heterogeneous data; Models, Statistical; Neural gas; Neural networks; Neural Networks (Computer); Training schedule
• ISSN: 0893-6080; 1879-2782
• DOI: 10.1016/j.neunet.2005.06.011

Data pruning and ordered training are two methods and the results of a small theory that attempts to formalize neural network training with heterogeneous data. Data pruning is a simple process that attempts to remove noisy data. Ordered training is a more complex method that partitions the data into a number of categories and assigns training times to those assuming that data size and training time have a polynomial relation. Both methods derive from a set of premises that form the ‘axiomatic’ basis of our theory. Both methods have been applied to a time-delay neural network—which is one of the main learners in Microsoft's Tablet PC handwriting recognition system. Their effect is presented in this paper along with a rough estimate of their effect on the overall multi-learner system. The handwriting data and the chosen language are Italian. 1 1 An abbreviated version of some portions of this article appeared in Drakopoulos and Abdulkader, 2005, published under the IEEE copyright.