Archives

2018

June 13th, 2018 Episode #10 Season #5

Meetup 5.10

June 13th, 2018

May 9th, 2018 Episode #9 Season #5

Meetup 5.9

May 9th, 2018

April 19th, 2018 Episode #9 Season #5

Hors Serie Devoxx

April 19th, 2018

Syllabus:

Comme tous les ans, on sera présent à la conférence Devoxx.

Les BOF (Bird of a feather) sont des sessions phares de Devoxx France, à partir de 19h45 le jeudi soir.

C'est le point de rendez-vous des user-groups, des communautés, des meetups, bref de tous les groupes communautaires de passionnés.

Passez d'une session du ParisJS pour échanger ensuite avec les membres du Paris Scala User Group, les BOFs se déroulent le soir de la soirée Meet-and-Greet.

Liste des B.O.Fs (Lien)

C'est une soirée d'échanges. On ne prévoit pas - pour le moment - un agenda précis (speaker, talk, etc).

L'inscription est *** OBLIGATOIRE *** 

password : FR2018

https://www.eventbrite.com/e/inscription-venez-decouvrir-devoxx-france-2018-le-temps-dune-soiree-44757308314

 

========= Important message from the conference's organiser =========

Hello,   As you maybe know, the BOF evening is traditionally held on the occasion of Meet & Greet (http://www.devoxx.fr/meet-and-greet). In order to allow you to invite to your BOF participants who are not registered to Devoxx France 2018 we offer you the opportunity to invite them before the official opening which will be done soon. 

For this they can register on the following eventbrite with the password: FR2018

https://www.eventbrite.com/e/inscription-venez-decouvrir-devoxx-france-2018-le-temps-dune-soiree-44757308314  

Do not forget that this year, the BOFs rooms are only able to accommodate about thirty participants and the people registered to Devoxx France do not need to register for the Meet & Greet. 

Best regards

April 11th, 2018 Episode #8 Season #5

Paris Machine Learning #8 Season 5, Funding AI, Deep GPU, Robot Cars, Industry

April 11th, 2018

Syllabus:

 

 

Check this awesome line-up we have made for you ! 

But .... We don't have a sponsor for the drinks and food !

 

Please contact Franck (https://www.linkedin.com/in/franckbardol/ ) or Igor ( https://www.linkedin.com/in/IgorCarron/ )if you or your organization want to host the meetup.

 

Thanks to IPGG (Institut Pierre-Gilles de Genes) for hosting this meetup.Schedule:6:30 PM doors opening ; 6:45 PM : talks beginning ; 9:00 PM : talks ending10:00 PM : end

 

Talks :

--- Rossini Rodolpho, zeroth.ai, we fund the stuff that nobody else will

Talk about the AI investments we do at Zeroth

 

 --- Guillaume Barat, NVIDIA, NVIDIA updates - How to accelerate AI ?

NVIDIA will come back on GTC annoucements (GPU Technology Conference) and how to accelerate AI workload.

 

--- Tarin Ziyaee, Voyage

We're bringing self-driving cars to a retirement community (and city) like no other: The Villages, Florida. With 125,000 residents, 750 miles of road and 3 distinct downtowns, The Villages is a truly special place to live.

Lien article: New York Times

 

--- Pierre Gutierez, Scortex.io, Automating quality visual inspection using deep learning

Driven by Industry 4.0, Scortex deploys artificial intelligence at the heart of factories. 

 

We offer a smart visual inspection solution for quality control. Scortex turnkey platform enables manufacturing companies to automate their most complex inspection tasks. 

 

 In this talk, we’ll share Scortex experience on computer vision for visual inspection in factory environment. We will explain what are our current challenges and how we plan to solve them. 

 

Then, on a real use case example, we will discuss how we generate data through our own acquisition system and what are the advantages and drawbacks of this from the machine learning point of view. We will also discuss our labelling process as well as the leads we have to reduce the labelling efforts on our side. 

 

-- Wenjie ZHENG, Learning Low-rank Matrices Distributedly without Factorization

Learning low-rank matrices is a problem of great importance in statistics, machine learning, computer vision and recommender systems.

Because of its NP-hard nature, a principled approach is to solve its tightest convex relaxation: trace norm minimization.

Among various algorithms capable of solving this optimization is the Frank-Wolfe method, which is particularly suitable for high-dimensional matrices.

In preparation for the usage of distributed infrastructures to further accelerate the computation, this study aims at exploring the possibility of executing the Frank- Wolfe algorithm in a star network with the Bulk Synchronous Parallel (BSP) model and investigating its efficiency both theoretically and empirically.

 

March 14th, 2018 Episode #7 Season #5

Meetup 5.7

March 14th, 2018

March 14th, 2018 Episode #7 Season #5

Paris Machine Learning Meetup #7 Season 5, NLU, AI for HR, decentralized AI

March 14th, 2018

Syllabus:

***** Joseph Dureau, Snips NLU, an Open Source, Private by Design alternative to cloud-based solutions

As part of its mission to expand the use of privacy-preserving AI solutions, the Snips team has decided to fully open source its solution for Natural Language Understanding. Snips NLU is an alternative to all cloud-based NLU solutions powering chatbots or voice assistants: Dialogflow, Luis.ai, Recast, Amazon Lex, Wit.ai, Watson, etc. 

You can run it on the edge or on premises, thus avoiding giving away your user data to a third party service.

website, presentation

 

***** Erik Mathiesen, Octavia.ai, An AI Careers Advisor: Using Machine Learning to Predict Your Career Path

Octavia.ai specializes in smart solutions for recruitment. In this talk, I will describe how we use AI, and in particular Neural Networks and Deep Learning, to analyse and predict people’s career paths. Having analysed millions of CVs, our system can predict from a person’s CV what jobs are most likely to be next in the career path of that individual, as well as when the next job move is mostly likely to happen. 

By doing this, we enable companies to predict and find better candidates as well as forecast future hiring needs within an organisation. 

I will outline the technologies and techniques used in this application and give a few illustrative example of its usage.

website, presentation

 

***** Morgan Giraud, OpenMined

An open-source community focused on building technology to facilitate the decentralized ownership of data and intelligence.

presentation

 

**** Open Foods Facts

Mentor recruiting session

website

March 5th, 2018 Episode #6 Season #5

Paris Machine Learning, Hors serie #3, Machine Learning & Graph DB

March 5th, 2018

Syllabus:

Détails

This meetup is the french part of the graphTour (https://neo4j.com/graphtour).

It is co-organised by neo4j and hosted by criteo. 

We thank criteo for hosting us and providing food & beverage.

 

Schedule : 

6:30 PM doors open ; 7 - 9 PM talks ; 9 - 10 PM networking ; 10PM end 

 

Talks :

Vlasta Kus (http://twitter.com/vlastakus), Graph-Powered Machine Learning

Graph-based machine learning is becoming an important trend in Artificial Intelligence, transcending a lot of other techniques. 

Using graphs as a basic representation of data for ML purposes has several advantages:

- the data is already modelled for further analysis, explicitly representing connections and relationships between things and concepts

- graphs can easily combine multiple sources into a single graph representation and learn over them, creating Knowledge Graphs;

- improving computation performances and quality. The talk will present these advantages and present applications in the context of recommendation engines and natural language processing.

 

Christophe (http://twitter.com/@ikwattro) (@ikwattro), 

Knowledge GraphLes Knowledge Graph, popularisés par Google, sont devenus une partie primordiale dans les processus d'aggrégation des données, de classification de documents et compréhension de langage. 

Neo4j est la plateforme de graphes numéro 1 qui vous permettra d'enregistrer et requêter ces données largement connectées.

Christophe vous décrira une approche basée sur les graphes pour de l'informatique cognitive et vous détaillera ce processus depuis l'ingestion de documents au format texte non-structuré jusqu'à la génération d'un Knowledge Graph, requêtable par language naturel via l'interface Alexa d'Amazon.

February 14th, 2018 Episode #6 Season #5

Paris ML #6 S5: Nano-neurons, Event extraction, Drug design, Health Hackathon

February 14th, 2018

Syllabus:

 

Thanks to Zenika for hosting us and sponsoring the event !

 

Schedule : 6:30PM doors open - 6:45PM to 9PM talks - 9PM to 10PM networking 

(first come first served basis then doors close)

  • Julie Grollier : neuromorphic architectures for AI
  • Marc Lelarge : winner of the Datathon MIT ICU APHP
  • Quentin Perron : Drug design
  • Serge Nakache / Hans Blick : Event extraction with Twitter API

===================

Details

+ Winning team Datathon MIT & APHP 2018 (http://blogs.aphp.fr/dat-icu/), Marc Lelarge **

General patient representation from electronic health records

We propose a deep learning algorithm to learn a low dimensional representation (embedding) of patients from their raw electronic health records. We evaluate our embeddings with descriptive analysis and a code assignement task. These representations would be reusable for other tasks such as patient similarity for cohort selection, information retrieval, electronic phenotyping, prediction.

 

+ Artificial intelligence for new drug design, Quentin Perron, iktos.ai (http://www.iktos.ai) **

New drug design is a long (5 years), costly (50-100M$) and unproductive process (1% success rate from hit to pre-clinical candidate)…

Iktos aims to leverage big data and AI to bring radical improvement to this process. Iktos has invented and is developing a truly innovative and disruptive artificial intelligence technology for ligand-based de novo drug design, focusing on multi parametric optimization (MPO). Our proprietary technology is built upon the latest developments in deep learning algorithms.

In a few hours, our technology can design new, druggable and synthesizable molecules, that are optimized to match all your selection criteria.

 

+ Artificial hardware nano-neurons, Julie Grollier, CNRS **

In this talk, I will show that magnetic nano-oscillators are promising building block for accelerating deep networks in hardware

Nano neurons in 2 min video :

 

+ Serge Nakache, Event detection with Twitter API

 

 

January 10th, 2018 Episode #5 Season #5

Meetup 5.5

January 10th, 2018