Presented By O'Reilly and Cloudera
Make Data Work
22–23 May 2017: Training
23–25 May 2017: Tutorials & Conference
London, UK

Faster deep learning solutions from training to inference

Nir Lotan (Intel), Barak Rozenwax (Intel)
14:5515:35 Thursday, 25 May 2017
Data science and advanced analytics
Location: Hall S21/23 (A)
Secondary topics:  Deep learning
Level: Beginner
Average rating: *****
(5.00, 3 ratings)

Who is this presentation for?

  • Data scientists and software developers

Prerequisite knowledge

  • Basic familiarity with machine learning

What you'll learn

  • Learn how to easily train and deploy deep learning models for image and text analysis problems using Intel's Deep Learning SDK

Description

Deep learning is the most significant innovation in data science in recent years, since it presents amazing improvements in modeling results. However, most data scientists don’t yet use deep learning, due to the relative complexity of customizing deep learning models for their own problems, the challenges in installing and using the required frameworks, and the low performance of open source deep learning frameworks on standard CPUs.

Barak Rozenwax and Nir Lotan explain how to easily train and deploy deep learning models for image and text analysis problems using Intel’s Deep Learning SDK, which enables you to use deep learning frameworks that were optimized to run fast on regular CPUs, including Caffe and TensorFlow. Intel’s Deep Learning SDK includes tools for easy data preprocessing and model compression, providing an end-to-end solution. In addition, it supports scale-out on multiple computers for training, as well as using compression methods and optimization for deployment of the models on various Intel hardware platforms, including those with limited memory footprints and performance constraints.

Barak and Nir walk you through a full demo of the deep learning workflow, from data preparation through training and compression to deployment for inference on various hardware platforms, and review the architecture used for scaling out existing deep learning frameworks across multiple CPUs for faster model training.

Photo of Nir Lotan

Nir Lotan

Intel

Nir Lotan is a machine-learning product manager and team manager in Intel’s Advanced Analytics department. Nir’s team develops machine-learning and deep learning-related tools, including a tool that enables easy creation of deep learning models. Prior to this role, Nir held several product, system, and software management positions within Intel’s Design Center organization and other leading companies. Nir has 15 years of experience in software and systems engineering, products, and management. He holds a BSc degree in computer engineering from the Technion Institute of Technology.

Photo of Barak Rozenwax

Barak Rozenwax

Intel

Barak Rozenwax is a machine-learning product owner and CSPO in Intel’s Advanced Analytics department, where he is part of a team that develops a deep learning training tool that enables easy creation and training of deep learning models. Barak’s previous roles included several product and system positions within his department in Intel. Barak has more than seven years of experience in software and systems engineering. He holds a BSc in industrial engineering and management with a focus on information systems from Ben-Gurion University of the Negev.