Node-By-Node Greedy Deep Learning for Interpretable Features
This work addresses the need for faster and more interpretable deep learning methods, though it appears incremental as it builds on existing sequential layer training approaches.
The authors tackled the problem of training deep networks more efficiently and interpretably by introducing a node-by-node greedy training algorithm, which is orders of magnitude faster and maintains out-of-sample performance while improving node-level interpretability.
Multilayer networks have seen a resurgence under the umbrella of deep learning. Current deep learning algorithms train the layers of the network sequentially, improving algorithmic performance as well as providing some regularization. We present a new training algorithm for deep networks which trains \emph{each node in the network} sequentially. Our algorithm is orders of magnitude faster, creates more interpretable internal representations at the node level, while not sacrificing on the ultimate out-of-sample performance.