Reconciling modern machine learning and the bias variance trade off

Use Tensorflow to test the bias-variance myth. Complete code walk through and tests of super-fitting a model to data.

In this walkthrough we create a neural network that leans to predict MNSIT as a baseline, then create a neural network that can superfit the data and we show that it still preforms well, dispelling the old held-to belief that too many free parameters will overfit.

Try out the colab notebook

Bias-Variance,Tensorflow,Python,Reconciling modern bias variance trade off,Anton,Machine Learning,learning tensorflow,arxiv paper,Keras,