Reading questions

1. Why did they use the linear model for regulation?  Describe pros and cons.

2. Describe the new ML idea this paper proposes.

3. Any room for improvement, in terms of modeling or learning?