Machine Learning
See also
Course material,
Other LABS,
Course projects
LABS
10. Deep Learning
We consider again the handwritten digits images available in the MNIST database. The goal here is to build a classifier to classify each (new) image in one of the classes 0, 1, ..., 9.
1. Fit a Deep Neural Network using the Keras module.
Try different architectures, activation fonctions, ....
2. Fit a CNN using the Keras module. (WARNING: the computational time may be long!)
Try different architectures, activation fonctions, ....
Codes:
digits_CNN.R,
digits_CNN.py,
digits_DeepNN.R,
digits_DeepNN.py,
Jupyter notebook example for keras