Wednesday, December 11, 2019

Ce Soir: Paris Machine Learning Meetup #2 Season 7: Symbolic maths, Data Generation thru GAN, "Prevision Retards" @SNCF, Retail and AI, Rapids.ai Leveraging GPUs

** Nuit Blanche is now on Twitter: @NuitBlog **


A big thank you to Scaleway for hosting us in their inspiring office and sponsoring the networking event afterwards.


So this is quite exciting. Our meetup group has 7 999 members and we are going to organize a meetup in a town that is paralyzed by strikes. During the course of existence of this meetup, we have seen worse.  

For those of you who will not be able to make it, all information slides and link to streaming are below:


Tabular data are the most common within companies. Generating synthetic data that respects the statistical properties of the original data can have several applications: a machine learning that respects data privacy, improving the robustness of a model in relation to data drift, etc. Since 2018, there has been an increasing number of academic publications presenting the use of GANs on this type of data, particularly on patient medical data. We have performed a proof of concept on real data, and present the results of several models from the research, namely the Wasserstein GAN, the Wasserstein GAN with Gradient Penalty and the Cramér-GAN, with the objective of "model compatibility", i.e. the possibility of using synthetic data to replace real data to train a classifier.

2. Eloïse Nonne, Soumaya Ihihi, "Prévisions Retards" a Machine Learning project led by e.SNCF's Data IoT team.
Its goal is to integrate predictions of train delays into the SNCF mobile application. Every day, our model predicts delays for the next 7 days, at each stop, for every train in Paris area network. The challenge of this project is to improve the reliability of passenger information and to provide more relevant routes for the application users. We will present the project, from the definition of needs and exploratory data analysis, to its industrialization in the cloud and the reliability of its predictions.

This talk is focussed on AI and ML applications in retail. Discover how Carrefour is transforming through the introduction of the Google - Carrefour Lab by Elina Ashkinazi-Ildis, Director of the Lab. Then go further with the "shelf out detection" usecase presented by Kasra Mansouri, Data Scientist within Artefact.

RAPIDS makes it possible to have end-to-end data science pipelines run entirely on GPU architecture. It capitalizes on the parallelization capabilities of GPUs to accelerate data preprocessing pipelines, with a pandas-like dataframe syntax. GPU-optimized versions of scikit-learn algorithms are available, and RAPIDS also integrates with major deep learning frameworks.
This talk will present RAPIDS and its capabilities, and how to integrate it in your pipelines.


Neural networks have a reputation for being better at solving statistical or approximate problems than at performing calculations or working with symbolic data. In this paper, we show that they can be **surprisingly good** at more elaborated tasks in mathematics, such as symbolic integration and solving differential equations. We propose a syntax for representing mathematical problems, and methods for generating large datasets that can be used to train sequence-to-sequence models. We achieve results that outperform commercial Computer Algebra Systems such as Matlab or Mathematica.
https://arxiv.org/abs/1912.01412



Follow @NuitBlog or join the CompressiveSensing Reddit, the Facebook page, the Compressive Sensing group on LinkedIn  or the Advanced Matrix Factorization group on LinkedIn

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.

Other links:
Paris Machine LearningMeetup.com||@Archives||LinkedIn||Facebook|| @ParisMLGroup
About LightOnNewsletter ||@LightOnIO|| on LinkedIn || on CrunchBase || our Blog
About myselfLightOn || Google Scholar || LinkedIn ||@IgorCarron ||Homepage||ArXiv

No comments:

Printfriendly