Genetically Evolved Extreme Learning Machine for Letter Recognition Dataset

摘要

It is well known that the performance of learning feed forward neural networks is in general far slower than required and it has been a major bottleneck in their applications. Two key obstacles the slow gradient-based learning algorithms which are extensively used to train neural networks. Combining slow training process with even slower evolutional methods appears to be incomprehensible but here comes the Extreme Learning Machine. ELM has randomly chosen hidden nodes and analytically determined only the output weights of network. In theory, this algorithm tends to provide good generalization performance at extremely fast learning speed. Experiment in this paper shows that ELM's classification efficiency can be noticeably improved if its training is combined with Genetic Algorithm.