Skip to content

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

Notifications You must be signed in to change notification settings

skj092/Ann-Experiments

Repository files navigation

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.

About

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

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published