Open stoplime opened 6 years ago
https://arxiv.org/pdf/1408.5882v2.pdf http://www.wildml.com/2015/12/implementing-a-cnn-for-text-classification-in-tensorflow/ Use a Word2Vec to do unsupervised pre-processing on the sentences before running it through a CNN for best performance. It doesn't have to be Word2Vec I don't think cause fastText is a better version of Word2Vec. Maybe we could use that? https://github.com/facebookresearch/fastText
List out all the popular Techniques being used in NLP. Such examples would be like attention CNN replacing recurrent networks.