Some of the projects that we have made during our universuty courses. The first one is about using a Neural Network to classify images from the dataset fashion_mnist, avaiable on the UCI Machine Learning Repository website.
- Progetto.py : Download of the dataset fashion_mnist and normalization.
- Multi_1.py : Definition through Tensorflow of a Multi Layer Perceptron class (MLP), with a single hidden layer.
- Alg_S.py : Genetic Algorithm to optimize the hyperparameters of the neural neutwork (number of hidden neurons, minibatch size for training, learning rate).
- Multi_2.py : Same as Multi_1.py, but with 2 hidden layers.
- Alg_S.py : Same as Alg_S.py, but with 2 hidden layers.
The modules that are needed to run the code above are Tensorflow, Numpy, Matplotlib, Deap. As of now Tensorflow is only compatible with Python versions <= 3.7.0