Designing Transparent Neural Networks
Neural networks designed to be as transparent as linear and additive models.
Mar 2021Essay
Where does a linear model end and a neural network begin?
Designing Transparent Neural Networks
Neural networks designed to be as transparent as linear and additive models.
Mar 2021Essay
Generalized linear and additive models are well-established, interpretable approaches to supervised learning.
The essay connects these approaches to the building blocks of neural networks, and shows that it is possible to design neural networks that are just as transparent.
- Published March 4, 2021
- Kind Essay
