SLAC, Kavli 3rd Floor Conf. Room
Join us Friday, June 7th at 9:30 am on the Kavli third floor at SLAC (and on zoom!) for the last KIPAC Stats & ML Journal Club of the quarter. Justin Myles will be leading a discussion on training Recurrent Neural Networks. Find a link to the paper and his abstract below.
Professor Forcing: A New Algorithm for Training Recurrent Networks
Abstract: The Teacher Forcing algorithm trains recurrent networks by supplying observed sequence values as inputs during training and using the network's own one-step-ahead predictions to do multi-step sampling. We introduce the Professor Forcing algorithm, which uses adversarial domain adaptation to encourage the dynamics of the recurrent network to be the same when training the network and ...
We’ll discuss Professor Forcing, an algorithm for training recurrent neural networks for the purpose of sequential sampling. This network design has applications for computer-generated sequential media (where each generated datum depends on its predecessor) like computer handwriting generation.