Statistical Learning

Some thoughts about why deep learning is interesting.

How differential privacy can be used to help solve problems of overfitting in adaptive data analysis.
First post in a series on generalization in machine learning.