Presented By O’Reilly and Intel AI
Put AI to work
Sep 4-5, 2018: Training
Sep 5-7, 2018: Tutorials & Conference
San Francisco, CA

Deep learning on mobile: The how-to guide

Anirudh Koul (Microsoft)
11:55am-12:35pm Thursday, September 6, 2018
Secondary topics:  Deep Learning models, Edge computing and Hardware

Who is this presentation for?

  • Data scientists, mobile developers, software architects, and researchers

Prerequisite knowledge

  • A high-level understanding of deep learning

What you'll learn

  • Learn how to deploy deep learning architectures on mobile devices
  • Gain practical tips for developing apps from real-world scenarios
  • Become familiar with the ecosystem and platforms available for AI on smartphones

Description

Over the last few years, convolutional neural networks (CNN) have risen in popularity, especially in the area of computer vision. Anirudh Koul explains how to bring the power of convolutional neural networks and deep learning to memory- and power-constrained devices like smartphones, wearable devices, and drones.

Many mobile applications running on smartphones and wearable devices would potentially benefit from the new opportunities enabled by deep learning techniques. Local execution allows data to stay on the mobile device, avoiding latency issues entailed by data transmission to the cloud and alleviating privacy concerns. However, CNNs are by nature computationally and memory intensive, making them challenging to deploy on a mobile device. Anirudh shares various strategies to circumvent these obstacles and build mobile-friendly shallow CNN architectures that significantly reduce the memory footprint and therefore make them easier to store on a smartphone; he also explains how to use a family of model compression techniques to prune the network size for live image processing, enabling you to build a CNN version optimized for inference on mobile devices. Along the way, he outlines practical strategies to preprocess your data in a manner that makes the models more efficient in the real world.

Anirudh showcases these techniques using a real-world project and discusses tips and tricks, speed and accuracy trade-offs, and benchmarks on different hardware to demonstrate how to get started developing your own deep learning application suitable for deployment on storage- and power-constrained mobile devices. You can also apply similar techniques to make deep neural nets more efficient when deploying in a regular cloud-based production environment, thus reducing the number of GPUs required and optimizing on cost.

Photo of Anirudh Koul

Anirudh Koul

Microsoft

Anirudh Koul is a senior data scientist at Microsoft Research and the founder of Seeing AI, a talking camera app for the blind community. Anirudh brings over a decade of production-oriented applied research experience on petabyte-scale datasets, with features shipped to about a billion people. An entrepreneur at heart, he has run ministartup teams within Microsoft, prototyping ideas using computer vision and deep learning techniques for augmented reality, productivity, and accessibility and building tools for communities with visual, hearing, and mobility impairments. A regular at hackathons, Anirudh has won close to three dozen awards, including top-three finishes for four years consecutively in the world’s largest private hackathon, with 18,000 participants. Some of his recent work, which IEEE has called “life changing,” has been showcased at a White House event, on Netflix, and in National Geographic and received awards from the American Foundation for the Blind and Mobile World Congress.