For a second year, Duke and Nvidia are sponsoring a GPU workshop, this time with a focus on deep learning technologies and frameworks Caffe, Theano, and Torch.

Registration is limited and required, and lunch will be provided. The session will take place on November 7, 2017, from 11:30-4:30 in the Technology Engagement Center (TEC, formerly known as the monumental Telcom Building).


  • Deep Learning Demystified
    Lecture introduces key terminology, use cases from various industries, how deep learning differs from previous algorithmic approach, and how a deep neural network gets trained, optimized, and deployed.
  • Applied Deep Learning
    Lecture covers how to apply deep learning to challenging problems, what types of problems benefit most from deep learning, what skills and knowledge is needed to use deep learning, and the characteristics of successful deep learning projects.
  • Labs
    Applications of Deep Learning with Caffe, Theano, and Torch

Frameworks: Caffe, Theano, Torch

This lab introduces the rapidly developing technology of deep learning accelerated by GPUs. The course is intended for anyone looking for a fundamental understanding of deep learning.
In this lab, you will learn:

  • The concept of deep learning
  • How the growth of deep learning has improved machine perception tasks including visual perception, speech recognition, and natural language
  • How to choose which software framework best suits your needs
  • Image Classification with NVIDIA DIGITS

Frameworks: Caffe

This lab shows you how to leverage deep neural networks (DNN) – specifically convolutional neural networks (CNN) – within the deep learning workflow to solve a real-world image classification problem using NVIDIA DIGITS on top of the Caffe framework and the MNIST hand-written digits dataset.
In this lab, you will learn how to:

  • Architect a Deep Neural Network to run on a GPU
  • Manage the process of data preparation, model definition, model training and troubleshooting
  • Use validation data to test and try different strategies for improving model performance