Exercises

  • Implement three different functions, each returning a network varying in size (depth and width). Use each of these functions and perform a k-fold cross validation. Assess which size fits best.
  • Experiment with MAE and MSE loss functions, and note the difference during training.
  • Experiment with different loss functions and note the differences during training.
  • Experiment with different regularization techniques and note the differences during training.