Deploying neural networks
for efficient inferencing
in edge devices
Hosted by Stephen Alderman
25 January, 2018
Neural networks (NN) are emerging as a key disruptive technology in a wide range of segments. They enable face recognition and scene understanding in smart cameras and enhanced user experiences in mobile devices through computational photography, photo management, speech recognition and synthesis, to name a few applications.
While many machine learning frameworks and desktop-based solutions exist today for the development and training of NNs, deploying NNs in edge devices is particularly challenging due to the high performance requirements and the limited resources available.
This webinar summarizes NNs in edge devices, including applications and trends. It then focuses on the key deployment challenges, the tools and techniques to deliver deployable and efficient NN solutions and how they are enabled by innovative new hardware acceleration technologies.
Benefits of watching:
- Understand NN market trends and applications
- Learn the key challenges, tools and techniques for deploying NN models for efficient inferencing in mobile and embedded devices
- Understand the process and benefits of optimising floating point NN models to fixed point with flexible bit depth
- Discover how new hardware acceleration technologies like the PowerVR 2NX Neural Network Accelerator can deliver the needed solution to meet performance and power requirements while integrating to industry standard machine learning frameworks and providing an easy-to-use workflow
About the speaker
Stephen Alderman, Business Development Executive for the PowerVR Vision & AI, Imagination Technologies
Stephen is Business Development Executive for the PowerVR Vision & AI at Imagination Technologies. Stephen joined imagination in 2011 as a placement student, and has continued to work at Imagination since completing his master’s degree in Electronic Systems Engineering.