README.md

Use ML-FLOW and TensorFlow2.0(Keras) to record all the experiments on Fashion MNIST dataset.

1. Study and document the evolution of ANN from Perceptron with derivation.
2. Document problems generally faced in training an ANN and their solution provided in the lecture with derivation.
3. Compare and document different available activation functions.
4. Compare and document different available weight initialization techniques.
5. Observe and document results before and after applying Batch Normalisation.
6. Observe and document results before and after applying Transfer Learning.
7. Observe and document use of Early Stopping and Check-pointing.
8. Compare and document different available Optimizers and their derivation.
9. Observe and document the use of various loss functions.
10. Observe and document results before and after applying various regularisation techniques like l1, l2, and dropout techniques.

----------------------------------------------------------------------\

Note : Some of the notebook contain latex code which doesn not support in github. You can open the notebook in google drive.