Classifying Glyphs by Combining Evolution and Learning

摘要

Artificial neural networks are used to classify the writing system of an unseen glyph. The complexity of the problem necessitates a large network, which hampers the training of the weights. Three hybrid algorithms - combining evolution and back-propagation learning - are compared to the standard back-propagation algorithm. The results indicate that pure back-propagation is preferable to any of the hybrid algorithms. Back-propagation had both the best classification results and the fastest runtime, in addition to the least complex implementation.