ImageTraining-A4

Prerequisites

Required Packages: (fetch with pip install)

keras
tensorflow
jupyter-notebook
scikit

FNIST.ipynb

Purpose:

Experiment with training several binary classifiers in just a few clicks. You can tweak the model’s architecture and hyperparameters to get some understanding on how neural networks work.

TensorFlow Playground: https://playground.tensorflow.org/

1. Try training the neural network by clicking the run button (top left).
2. Try replacing the tanh activation function with the reLU function and train the network again.
3. Modify the network architecture to have just one hidden layer with three neurons.
   Train it multiple times (to reset the network weights, click the reset button).
4. Remove one neuron to keep just two. Retrain the network.
5. Set the number of neurons to eight and train the network several times.

MNIST.ipynb (WIP)

Important:

Load data with keras.datasets.mnist.load_data()

We will train a deep multilayer Neural Network on the MNIST dataset. The MNIST dataset is a set of images of handwritten digits 0-9, with associated target casses numbered as 0-9.

We will play around with different layers and hyperparameters. Then, we'll train the model on the training set before testing it on the test set. In this particular example, the splits are already done for us ahead of time.

Changes