DeepMola Neural Network [DML]

This library is built on top of theano.

Requires:

DML use theano and scipy, and run with python 3

• python3
• numpy
• skimage
• theano (do not forget theano dependencies)
• matplotlib for plotting network training accuracy

How to test examples

To launch an example of the example folder, you should use:

python3 main.py <example_name>

For example, you can launch the mnist example by typing python3 main.py mnist. Before starting the example, you download the dataset in the data folder.

Current examples:

• mnist

How to use the library

• Create a model within you python code. After creating the object, you can still modify it, add layers, checkers, etc...
• Build the model (build function). After this operation, do not update change the model without re-building
• Train your model with the wanted parameters (train function)
• You can then use you model to predict, classify, or whatever you want, using runSingleEntry or runBatch methods.