Keras-NN-Tutorial

Learn to compile, train, and evaluate a neural network using the Keras machine learning framework

Setup

Clone the repository:

git clone https://gitlab.cern.ch/ou-itk-pixels/keras-nn-tutorial.git

We will use a conda environment to manage the different machine learning packages that will be used. You can install conda for your OS here: miniconda

Now, we will create the conda environment named ml-env and install some packages.

conda create -n ml-env
conda activate ml-env
conda install tensorflow
conda install keras
conda install scikit-learn
conda install pyyaml
conda install notebook
conda install pandas
conda install matplotlib
conda install seaborn

The environment can be exported to a text file with:

conda list -e > requirements.text

Then a fresh enviroment can be created from this requirements.txt by doing:

conda create -n ml-env --file requirements.txt

Activate the environment with all the packages installed and then launch a jupyter notebook

conda activate ml-env
jupyter notebook

Open keras_nn_tutorial.ipynb and work through the tutorial