The Future of Model Efficiency for Edge AI

On-Demand Webinar

The Future of Model Efficiency for Edge AI

AI model efficiency is crucial for making AI ubiquitous, leading to smarter devices and enhanced lives. Besides the performance benefit, quantized neural networks also increase power efficiency for two reasons: reduced memory access costs and increased compute efficiency.

The quantization work done by the Qualcomm AI Research team is crucial in implementing machine learning algorithms on low-power edge devices. In network quantization, we focus on both pushing the state-of-the-art (SOTA) in compression and making quantized inference as easy to access as possible. For example, our SOTA work on oscillations in quantization-aware training that push the boundaries of what is possible with INT4 quantization. Furthermore, for ease of deployment, the integer formats such as INT16 and INT8 give comparable performance to floating point, i.e., FP16 and FP8, but have significantly better performance-per-watt performance. Researchers and developers can make use of this quantization research to successfully optimize and deploy their models across devices with open-sourced tools like AI Model Efficiency Toolkit (AIMET).

In this webinar you will learn about:

  • The state-of-the-art in AI model efficiency from Qualcomm AI Research’s latest papers
  • How 4-bit integer weight quantization is possible without sacrificing much accuracy using our advanced Quantization-Aware-Training (QAT) techniques
  • The benefits of quantization and why we recommend integer inference (INT4, INT8, INT16) over floating point inference (FP8, FP16, FP32)
  • How Qualcomm’s AI research in quantization is feasible in real-life applications
  • The tools available for AI developers to implement their models on low-power edge devices: AIMET and the AIMET Model Zoo

Overview

September 17th: Day 1 – Keynote & Presentations 8:00am-4:00pm
Location: Qualcomm – N Auditorium 5775 Morehouse Drive, San Diego, CA

September 17th: 4:30pm-6:30pm: Networking Meetings & Reception
Location: La Jolla Marriott 4240 La Jolla Drive, San Diego, CA

September 18th: Day 2: 8:30am – 12:30pm Presentations & Meetings
Location: La Jolla Mariott 4240 La Jolla Drive, San Diego, CA

Highlights

Opening Keynote:
Qualcomm EVP, Engineering & CTO Jim Thompson

The Internet of Things – Challenges and Opportunities:
Qualcomm Technologies SVP & GM, IoT Jeff Lorbeck

Industry 4.0/5G and Smart Cities:
Qualcomm Technologies VP, Business Development Jeffery Torrance

Qualcomm Smart Cities Accelerator Program:
Qualcomm Technologies Sr. Director, Business Development, Head of Smart Cities Sanjeet Pandit
Industry Panel:
Smart Cities Deployments: Experiences and Learnings

Industry Panel:
5G/Industry 4.0/Auto: Future Smart Spaces

City Panel:
Use Cases

Speakers

Chirag Patel
Chirag Patel

Principal Engr./Mgr. in the Corp. R&D AI Research team

Chirag Patel is a Principal Engr./Mgr. in the Corp. R&D AI Research team at Qualcomm Technologies, Inc. As AI Model Efficiency Toolkit (AIMET) project lead, he is responsible for bringing neural network model efficiency R&D to practice working with inter-disciplinary teams, feature roadmap planning, and customer engagements. He also leads projects for enriching smartphone UX experience using a combination of machine learning and low power, always-on sensors, and forward-looking technologies. He has 10+ years of experience in wireless communications as a research engineer, leading design & standardization of standalone LTE in unlicensed spectrum and 3G/4G small cell technologies. He holds a Ph.D. in Electrical Engineering from Georgia Institute of Technology, Atlanta.

Tijmen Blankevoort
Tijmen Blankevoort

Director of Engineering at Qualcomm Technologies Netherlands BV

Tijmen Blankevoort is a Director of Engineering at Qualcomm Technologies Netherlands BV. Tijmen Blankevoort is the team lead for compression and quantization research at Qualcomm Technologies Netherlands B.V. With a background in mathematics and artificial intelligence, he started a deep learning start-up in 2013 with Professor Max Welling, which was later acquired by Qualcomm Technologies in 2017. The compression and quantization research team focuses on making neural network models more efficient, ensuring that low-bit quantization can be achieved through an automatic process without sacrificing much accuracy. Tijmen and his team are conducting new research in this area, and simultaneously bridging the gap between research and practice. In his spare time, Tijmen loves to play Magic: The Gathering, and is a fervent molecular gastronomy cook.


Speaker 4 Name
Speaker name

Job title
Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.


...
Card title

Some quick example text to build on the card title and make up the bulk of the card's content.

Have a question about this event?

Email: [email protected]