Wednesday, March 09, 2016

Ce Soir Paris Machine Learning Meetup #10 Season 3, M&A, Speech Recognition, Telecoms, Mariana DL & HR

Video of the streaming is below:

If you are waiting and want to watch there is the streaming of Match 1 of the Google DeepMind Challenge Match: Lee Sedol vs AlphaGo here.

The meetup is sponsorised by Capgemini Consulting (room and buffet). A big thank you to them.

Program of the meetup with presentation slides/ Le programme du meetup :

Talk : 19h - 21h:30

Franck Bardol, Igor Carron, Introduction

Olivier Auliard et Charlotte Delaunay (Capgemini Consulting) Mergers and Acquisitions

Savoir anticiper les fusions et acquisitions est un enjeu particulièrement crucial pour les banques d'affaires. A la demande de l'une d'entre elles, nous avons construit un modèle prédictif permettant d'identifier quelques mois à l'avance les entreprises susceptibles de devenir une cible pour une fusion ou une acquisition. Nous proposons de vous détailler la démarche que nous avons mise en oeuvre, depuis sa conception jusqu'aux éléments de visualisation des résultats obtenus, en passant par l'acquisition des données, leur préparation, la construction de la cible, les modèles testés...

- Gabriel Synnaeve (FAIR)
Summary of his talk @ Collège de France :
Deep Learning in Speech Recognition : we show some examples of the impact of deep learning on the whole field of speech recognition. We will also give a quick glance at one of the state of the art systems.

- Stéphane Senecal (Orange)
 Reinforcement Learning for Resource Allocation in Wireless Communication Networks

The works presented in this talk address a resource allocation problem arising in the context of mobile communication networks. More specifically, this problem deals with the optimization of network energy consumption for a guaranteed quality of service level. The solution investigated is based on reinforcement learning schemes for implementing resource allocation tasks for the network. The problem of interest can indeed be cast into a Markov Decision Process (MDP) framework. It appears from the experiments that state-of-the-art controller systems can be improved, so that an alternative MDP modeling is proposed and its performance results are compared to the state-of-the-art, demonstrating significant performance improvement over the latter. 

- Tariq Daouda (University of Montreal) Mariana, the Cutest Deep Learning Framework
Talk around Mariana Deep Learning Library :
As neural nets increase in complexity they also become harder to write and harder to teach. Our hypothesis is that these difficulties stem from the absence of a language that elegantly describe neural networks. Mariana (named after the deepest place on earth, the Mariana trench) is an attempt to create such a language within python. That being said, you can also call it as an Extendable Python Machine Learning Framework build on top of Theano that focuses on ease of use.
Tariq viens de créer un dossier publique qui contient ses présentations sur Mariana, il y a inclut les slides de la présentation du meetup ainsi que l'exemple de live coding.
- Martin Prillard (

Slides pdf 

Comment la psychométrie et le machine learning peuvent identifier les cultures d'entreprise et les facteurs de réussite professionnels

- Buffet - Networking : 21h30 - 22h30

Join the CompressiveSensing subreddit or the Google+ Community or the Facebook page and post there !
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.

No comments: