/NeuralNets

Some experiments on neural networks and deep learning

Primary LanguageJupyter NotebookOtherNOASSERTION

Some experiments on neural networks and deep learning

In this repository, some techniques for improving neural networks and deep learning are described, implemented and tried out.

Content

MNIST Experiments:

  • Use MC dropout for improving classification (mc_dropout.ipynb)

Regression uncertainty:

  • Train simple model on Boston Housing Price dataset and estimate uncertainties of predictions (UncertaintyEstimationRegression.ipynb)