Vikram Saletore of Intel will be joined by Lucas A. “Luke” Wilson of Dell EMC, an Intel® AI Builders partner, to discuss large-scale neural network training on Intel® Xeon® Scalable processors.
Luke Wilson will present tests that were performed at Dell EMC on CheXNet, a Stanford University project which extends a DenseNet model pre-trained on the large-scale ImageNet dataset to detect pathologies in chest x-ray images, including pneumonia and emphysema. In particular, the session will examine the improved time to solution on extended training of this pre-trained model, as well as whether various storage and interconnect options lead to more efficient scaling. Luke will outline the process that allows the user to train a model in a matter of minutes, look at the accuracy of the results, then re-test with a new set of hyper parameters until the model converges with the accuracy desired to deploy in the real world.
Can’t make it to the session? Check out the Efficient Neural Network Training on Dell EMC Intel Xeon® session at AIDC San Francisco earlier this year.
The Artificial Intelligence Conference, is co-presented by O’Reilly and Intel. For the full conference schedule and additional details, click here.
See the full session description here.