Presented By O’Reilly and Cloudera
Make Data Work
March 5–6, 2018: Training
March 6–8, 2018: Tutorials & Conference
San Jose, CA

Taming deep learning

Evan Sparks (Determined AI)
2:40pm–3:20pm Wednesday, March 7, 2018
Data engineering and architecture, Data science and machine learning
 Location: LL20 C
Average rating: *****
(5.00, 1 rating)

Who is this presentation for?

  • ML engineers, directors of data science, and executives considering investment in deep learning

Prerequisite knowledge

  • A basic understanding of machine learning and deep learning
  • Familiarity with existing deep learning application frameworks, such as TensorFlow, Keras, or Caffe (useful but not required)

What you'll learn

  • Understand the conventional workflow associated with building a deep learning model and shipping it in production
  • Learn key challenges in this workflow and how they can be addressed

Description

Deep learning has yielded a step function improvement on important problems ranging from computer vision to natural language processing, and there is enormous excitement about its broader potential. However, building practical applications powered by deep learning remains an enormous challenge: the necessary expertise is scarce, the hardware requirements can be prohibitive, and current software tools are immature and limited in scope.

Evan Sparks outlines the current state of deep learning application development and details several key challenges associated with standard workflows, such as the cost of human and compute resources, training time and the development cycle, data and cluster resource management, and several issues around application deployment. Drawing on academic research from CMU, Berkeley, and UCLA and his experience at Determined AI, a startup that provides software to make deep learning engineers fantastically more productive, Evan then shares potential solutions that can dramatically enhance the velocity of application development

Topics include:

  • A performance model-based approach to distributed resource management and sharing
  • Superhuman workflow automation via a state-of-the-art method for hyperparameter optimization
  • Robust versioning and tracking of experiment metadata
  • A tight coupling between application development and deployment to bridge the gap between research and production
Photo of Evan Sparks

Evan Sparks

Determined AI

Evan Sparks is a cofounder and CEO of Determined AI, a software company that makes machine learning engineers and data scientists fantastically more productive. Previously, Evan worked in quantitative finance and web intelligence. He holds a PhD in computer science from the University of California, Berkeley, where, as a member of the AMPLab, he contributed to the design and implementation of much of the large-scale machine learning ecosystem around Apache Spark, including MLlib and KeystoneML. He also holds an AB in computer science from Dartmouth College.