Hands-on tutorial on RNNLM toolkit

Опубликовано 12 августа 2016, 2:28

This tutorial is an extension of TomasΓÇÖs last presentation on ΓÇ£Language modeling with neural networksΓÇ¥. Tomas will walk through the toolkit with real examples and scripts, so people will have chance to test the toolkit on-site if bring a laptop. In his last presentation, Tomas discussed the state of the art techniques for statistical language modeling. This includes recurrent neural network based language model and extensions that allow efficient training on large corpora: class-based neural net language model and joint training with a maximum entropy model. Results will be presented on standard data sets for language modeling and speech recognition: the Penn Treebank corpus, Wall Street Journal ASR, and NIST RT04 broadcast news speech recognition.