December 2 tinyML meetup to be held in conjunction with the Arm AIoT Dev Summit

Note: There is no charge to attend this event, but we do ask that you complete our registration form:
https://fs24.formsite.com/meptec/200/index.html

• 5:30pm - 6:00pm Check in / food will be served

• 6:30-7:00 Talk#1 – Neural Network Optimizations for On-Device AI
Presenters: Lingchuan Meng, Principal Engineer, Arm / Naveen Suda, Principal Engineer, Arm

On-device AI brings unprecedented capabilities and opportunities to endpoint devices with improved privacy, security, and reliability. Deploying on-device AI must consider various constraints of memory, power, latency, and cost. Therefore, neural network optimizations become critical for exploiting hardware features, reducing memory footprint and improving computation efficiency.
In this talk, we present techniques for model optimizations such as pruning, clustering, quantization and algorithmic transforms for enabling energy-efficient AI on resource-constrained devices. Optimization cascading is achieved where each technique preserves the preceding attributes. Under different optimization objectives, exhaustive search for finding the optimal sparsity, quantization and clustering levels for each layer is compute intensive and impractical. To counter this, we have developed efficient techniques to navigate through the large search space, thus enabling trade-offs between accuracy, latency and compressibility.”

Presenter bios:
Lingchuan Meng is a Principal Software Engineer working on ML model optimization and deployment at Arm Machine Learning Group. He has also contributed to Arm’s NPU design by inventing new algorithms for activation compression and convolution kernels. Prior to joining Arm, he worked at Qualcomm Research where he developed the Snapdragon Math Libraries. Lingchuan received his PhD in Computer Science from Drexel University, Philadelphia, with his contributions to automatic code generation and algorithms for DSP and polynomial arithmetic.
Naveen Suda is a Principal Engineer in Machine Learning (ML) Technology Group at Arm San Jose, working on ML model optimization targeting embedded devices. He contributed to enabling ML on Arm Cortex-M microcontrollers through CMSIS-NN and demonstrated multiple ML use-cases on microcontrollers. He received his Ph.D. from Arizona State University, Tempe in 2015. Dr. Suda is an author of 21 research publications and inventor on 2 pending US patents. Prior to his Ph.D., he was a Circuit Design Engineer at IBM, Bangalore, working on server-class processor design. His current research interests include building energy-efficient ML solutions for the IoT edge.

• 7:00-7:30 Talk#2 – CMSIS-NN: Efficient Neural Network Kernels for Arm Cortex-M CPUs
Presenter: Felix Johnny, Sr. Engineer, Arm

Abstract:
Deep Neural Networks are becoming increasingly popular in always-on endpoint devices performing data analytics right at the source, reducing latency as well as energy consumption for data communication. We will introduce CMSIS-NN, efficient kernels developed to maximize the performance of neural network (NN) applications on Arm Cortex-M processors. We will also talk about how Arm is collaborating with Google TensorFlow Lite team to bring the enhanced machine smarts to the tiniest Arm MCUs.

Presenter Bio:
Felix Johnny maintains Arm’s open source CMSIS-NN library that targets optimized Neural Network kernels for Cortex-M CPUs. He has spent most of the last 15 years in the wireless domain working with optimizations in memory and cycle constrained systems. During this period at Ericsson and ARM, he has worked closely with HW IP designers in developing accelerators and custom microprocessors that are well mated to the software design.

• 7:30-7:50 Talk#3 – OpenMV
Presenter: Kwabena Agyeman, President, OpenMV

Abstract and bio to come.

Meetup to be held in conjunction with Arm AIoT Dev Summit in Computer History Museum, Mountain View, CA, 2nd floor, Hahn Auditorium