Investigation of Maxout Networks For Speech Recognition

We explore the use of maxout neuron in various aspects of acoustic modelling for large vocabulary speech recognition systems; including low-resource scenario and multilingual knowledge transfers. Through the experiments on voice search and short message dictation datasets, we found that maxout networks are around three times faster to train and offer lower or comparable word error rates on several tasks, when compared to the networks with logistic nonlinearity. We also present a detailed study of the maxout unit internal behaviour suggesting the use of different nonlinearities in different layers.