Deep Learning Workshop
A hands-on introduction to Deep Learning with self-correcting exercises.
This repository contains resources used to teach Deep Learning at the Ecole Nationale Supérieure de Cognitique.
Course slides, available in French.
Code samples and exercises, under the form of Jupyter notebooks.
You need to know the basics of programming with the Python langage to follow along. If you don't, there are numerous online resources for discovering Python. Among others, check out the official Python tutorial or take a whirlwind tour of Python.
How to use the notebooks
Browse the notebooks online with the Jupyter notebook viewer.
Launch an interactive session using Binder:
Click on any
.ipynbfile in the notebooks folder to show it through GitHub's own notebook viewer.
Clone or download this repository and run a Jupyter notebook server on your local machine. See below for detailed instructions.
Local installation with Anaconda
Anaconda is a popular Python data science platform. It includes many useful librairies and tools out of the box, including the Jupyter server needed to run notebooks locally.
Install the Anaconda distribution on your machine.
Clone this repository into your working directory with git. Alternatively, you can download it as a
.zipfile, then extract the downloaded archive to the desired folder.
Open the Anaconda prompt and navigate to the
deep-learning-workshopfolder created during the previous step.
Install the required packages for the notebooks to run by typing the following command.
conda env create -f environment.yml
This command will create a new environnement named
dl-workshop. It might take several minutes to complete. See the conda documentation for more information about environments.
Activate the newly created environment by typing the following command.
If you're on OS X or Linux, you have to type
source activate dl-workshopinstead.
If you're on Windows, define TensorFlow as Keras' backend engine with the following command.
Navigate to the
notebookssubdirectory of the
deep-learning-workshopfolder, then launch a Jupyter server with the following command:
The previous command should open a new tab in your browser showing the list of notebooks (if not, visit http://localhost:8888). Click on
index.ipynbto get started!
If you're on Windows and notebook execution gets very slow over time, close the notebook server and restore TensorFlow as backend.
Relaunch the server: you should witness a noticeable speedup in notebook execution speed.
The code in this repository, including notebooks, is released under the MIT license.
The text content is released under the Creative Commons BY-NC-SA license.