Fall 2013

Provably Efficient Algorithms for Training Deep Networks

Wednesday, October 23rd, 2013 2:15 pm2:45 pm

Add to Calendar

We present a provably efficient, layer-by-layer, algorithm for training deep sum-product neural networks. Our algorithm comes with formal polynomial time convergence guarantees. Similar to other deep architectures, the resulting network can compactly represent any function on a finite training set. We also discuss provably efficient algorithms for training other types of neural networks.