Grammarly AI Club Meet-up #20: Drawing a “map of words”: word embeddings and applications to machine translation and sentiment analysis (подія в архіві)

19 февраля 2015 (четверг)
Kyiv, Tereshchenkivska, 13, office 110, Grammarly

Dear friends, we are glad to announce the 20th AI Club Meet-up at Grammarly.

Please register to participate :)

Speaker: Mostapha Benhenda
Mostapha retired from academic mathematics (dynamical systems and ergodic theory), and now he is beginning in machine learning. Mostapha is from France, he would like to have a regular deep learning meet-up in Kyiv, with a technical focus, but open to motivated beginners.

Language: English

Abstract: A word embedding is a function that embeds words into a vector space, in a way that captures their meaning. This embedding is obtained by training a statistical model on large amounts of unlabeled text (it is an unsupervised learning method).
The most popular algorithm is Google’s Word2vec, but for pedagogical purposes, we present Stanford’s GloVe algorithm, which is easier.
We’ll also experiment with applications to machine translation, and to sentiment analysis of movie reviews.
Word embeddings are an example of representation/feature learning, which is an active field of research (around "deep learning",etc..) with fast-growing business applications, related to text processing, computer vision, speech recognition, marketing.

See you there!

До обраногоВ обраному0
Дозволені теги: blockquote, a, pre, code, ul, ol, li, b, i, del.
Ctrl + Enter
Дозволені теги: blockquote, a, pre, code, ul, ol, li, b, i, del.
Ctrl + Enter

Підписатись на коментарі