This February, we will have four meetups (on the 10th, 11th, 17th and 22nd)! Maybe it's a sign of times or maybe it's because there are 29 days this month, who knows. Let us note that we now have 3200 members and more than 1000 members on our LinkedIn group. Today, we will have the first of two meetups this week.
We will be hosted by Maltem Consulting Group who are also sponsoring the networking event afterwards. For this meetup we will have the following presentation (slides will be linked here before the actual meetup)
One short presentation of a project presented at the #Necmergitur:hackaton:
* Pitch; Manga Zossou, Projet AZTEC : Audio sensors for threat detection/système de capteurs audio pour détecter des menaces) at 11 minutres in the video. (in French)
* Franck Bardol and Igor Carron, introduction.
• Julie Josse, Beauty and danger of matrix completion methods: unveiling a black box's subtleties for better decision at 1h14 minutres in the video. (in French)
• Andrei Yigal Lopatenko, Head of Search Quality @ WalmartLabs, (remote from SF) What problems of ecommerce can deep learning solve? at 52 minutres in the video. (in English)
A short overview of ecommerce problem which can be solved with deep learning method with a tech dive into image similarity as a product recommendation problem.
• Alex Perrier, (remote from Boston) at 24 minutres in the video. (in French)
Topic modeling avec LSA, LDA et STM appliqué aux streams de followers Twitter.
Topic modeling: LSA, LDA, STM avec code en python et R, Text mining, comment determiner le nombre de topics, comment visualiser les topics.
• Jean-Marc Marty, Proxem, The Quest for Cross-Lingual Systems at 1h57 minutres in the video. (in French)
At Proxem, our clients ask us to extract information from e-mails, social medias, press articles, and basically any type of text you can imagine. In the standard case, the text to process is written in various languages. To establish systems that support a wide scale of languages and formats is one of the mission of our Research team.
I will focus during this talk on a paper that we've presented at EMNLP 2015 called Trans-Gram: Fast Cross-lingual Word Embeddings. The objective of this paper is to introduce a model that learns aligned word embeddings throughout a significant number of languages in a scalable way.
Liked this entry ? subscribe to Nuit Blanche's feed, there's more where that came from. You can also subscribe to Nuit Blanche by Email, explore the Big Picture in Compressive Sensing or the Matrix Factorization Jungle and join the conversations on compressive sensing, advanced matrix factorization and calibration issues on Linkedin.