TheEimer/DL_Lab

Exercises for the deep learning lab course 2017, Uni Freiburg

Exercises for the deep learning lab 2017

Exercise 1:

Implement a feed-forward net from scratch and train it on MNIST data

Exercise 2:

Use tensorflow to compare learning rates and the runtime of different numbers of filters on a CNN

Exercise 3:

Train a CNN to solve a labyrinth task using supervised learning. Check how well the agent deals with changing environments (spoiler: not well).

Exercise 4:

Use Q-Learning to try and improve generalization performance from exercise 3.