Speaker: Chris Auld, Principal Technical Evangelist Manager Microsoft
As deep learning becomes more and more popular organizations are finding opportunities to scale to lager models and larger datasets. Cloud computing provides the potential of large scale computing resources including GPUs but taking advantage of these can be a challenge due to the need to either distribute graph computation or manage the distributed update of network weights. In this session we will cover practical approaches to the training of Deep Neural Networks at scale. We will look at the support provided by various frameworks for distributed training covering the following open source tools TensorFlow with Horovod ChainerMN and CNTK.
Track: Artificial Intelligence
Room: Lecture Theatre
Date: Saturday, 24th March, 2018
Produced by Engineers.SG