As part of our outreach program, Criteo AI Lab is proud to offer the Machine Learning community, a Crash-course on Deep Learning. This workshop, free of charge, will be delivered by Aurélien Géron, author of Hands-On Machine Learning with Scikit-Learn and TensorFlow (O'Reilly Media).
It will be hands-on: 20-30 minutes of lectures, followed by 20-30 minutes of practical exercises with TensorFlow.
Audience: Daily practitioners of machine learning either for academic or industrial purposes. We also encourage our attendees to contribute/attend future instances of the TensorFlow meetups and be an active member in the Paris ML community.
Â
Course level: Advanced
Dates:Â June 20, 24 & 26 2019
Time: From 9:00 am to 6:00 pm
Duration: 3 full days
Â
Deliverables: Printed slides presentations, Jupyter notebooks.
Â
Language: EnglishÂ
Â
Location: Criteo - 32, rue Blanche - 75009 Paris
Â
Price: Free
Â
IMPORTANT
Considering the density of the course, participants are expected to dedicate some time between training days to progress on exercises/homework.
Â
Workshop attendance is by application only. Seats are limited. We will choose the participants through the application process detailed below.
Â
APPLICATION PROCESS
Â
Participation to the workshop will be determined based on the level of proficiency in machine learning. You will be asked to submit your resume or LinkedIn profile when you apply, to assess the said proficiency. Also, to attend the workshop, all participants must comply with the prerequisites.
Â
Please note that the deadline for the application to the workshop is June 7th, 6:00 PM CET.
Applicants will be notified about the outcome of selection process on June 12th EOD, by email.
Â
Aurélien Géron is a Data Science consultant and trainer, former lead of YouTube’s automatic video classification team at Google. During his 20-year career in I.T., he founded several successful companies, now employing over 300 people, including Wifirst (founder and CTO from 2002 to 2012), a leading Wireless ISP in France.
He also wrote multiple courses and published several technical books, including the book Hands-on Machine Learning with Scikit-Learn and TensorFlow (O’Reilly Media, April 2017), the #1 best-selling book on Amazon in the Machine Learning category in 2017-2018. Second edition to be released soon.
Basics of TensorFlow 2.0: operations, tensors, auto-differentiation, optimizers, graph function.
Simple example: linear regression with TF 2.0, from end to end.
Load and process data efficiently with TF Data.
Introduction to neural networks.
Use tf.keras to create a dense neural network.
Save and load a model.
Visualize learning curves with TensorBoard.
Techniques for training deep neural networks: Xavier / Glorot and He initialization, ReLU / ELU / SELU activation functions, Batch Normalization, Dropout, Data Augmentation, early stopping ...
Introduction to the TensorFlow architecture: execution engine, parallelism, kernels, distributed mode, build.
How to program in Python.
How to use the NumPy library.
The basics of linear algebra:Â
What is a vector, a matrix?
How to multiply and transpose them.
Â
Know the basics of Machine Learning:
What is Machine Learning?Â
What is a model?
What is a cost function?Â
What does it mean to train a model?Â
What is the difference between a model parameter and a hyper-parameter?Â
What does "regularizing a model" mean?Â
What is over-fitting?Â
What are the training, validation, and test sets?Â
What is a cross-validation?
Â
No laptop will be provided during the training. Therefore you must come with your own material.
Hardware requirements:
Recent model (less than 2 years, eg. MacBook Pro 2016 or newer edition) preferably with a TensorFlow-compatible GPU card (even if a GPU card is NOT mandatory to follow the course).
Software requirements:
CUDA Toolkit & cuDNN (if you have a TensorFlow-compatible GPU card in your laptop)
Python 3.5 or 3.6 (not 3.7)
Python librairies :Â
Scikit-LearnÂ
TensorFlow 2.0 preview (the 'GPU version' if you have a TensorFlow-compatible GPU card in your laptop) https://www.tensorflow.org/install/pip
SciPyÂ
NumPyÂ
MatplotLib
Jupyter
Each laptop must be fully equiped and tested prior to the workshop.
Â
Â