A Journey to the Tip of Neural Networks

tl;dr

The below Jupyter notebook starts with the most basic tools of machine learning: Linear regression and logistic regression. Then we can go beyond the linear prediction via the scheme of basis expansion using the same linear methods. The basis expansion model and the neural network model seem similar but also different.

In the analogy of parenting style, the basis expansion is the helicopter mom, while the neural network mom is hand-off parenting style. The conclusion is that the basis expansion model has clear interpretation, while the neural network model has black-box characteristic but flexibility.

Notebook for more

Below is my jupyter notebook (hosted on kyso.io) containing the whole story. It is originally uploaded on the Github repository.

References

• The Element of Statistical Learning, 2E, Hastie et al.
• https://github.com/dgkim5360/the-elements-of-statistical-learning-notebooks#(shameless plug)