Statistical Modeling and Parameter estimation
Mis à jour :
1. Introduction
Overview to supervised learning The missing piece BOW: drawbacks
2. Feature engineering for words
Standard features in NLP Feature engineering Promises of deep learning Word-embeddings
3. Feature engineering for documents
Distributional hypothesis
“A synopsis of linguistic theory” John Firth 1963 Harris substitutability theory Term-Document matrix Distributional hypothesis Co-occurrence matrix Weighting Co-occurrence matrix
4. Word2vec
Learn the representation Word2Vec [Mikolov et al. 2013] Word2Vec Fun with Word Embeddings Drawbacks of word embeddings
Laisser un commentaire