Accelerate ML Development With Cloud-Based Arm Cortex-M Models

Explore a 55-minute tinyML Talk that delves into accelerating machine learning development using cloud-based Arm Cortex-M models. Learn about Arm Virtual Hardware and its capabilities in providing accurate models of Cortex-M based processors for application developers. Discover how this technology integrates into desktop IDE-based development and cloud-hosted CI/CD and MLOps workflows. Understand the Virtual Streaming Interface (VSI) and its role in feeding real-world or synthetic data into ML applications for comprehensive testing and validation. Gain insights into how the TensorFlow OSS project utilizes Arm Virtual Hardware to enhance Arm Cortex-M based target optimizations. The talk covers various aspects including software stack, CMSIS, bundled software components, package building, test and configuration matrices, scaling, virtual streaming interface, event recording, performance analysis, and includes a demo and resources. Presented by Matthias Hertel, Product Specialist at Arm, this talk offers valuable knowledge for developers working with ML on embedded systems.

Intro
Software stack
cmsis
Bundled software components
Building a pack
What does it supply
Test matrix
Configuration matrix
Scaling
Real world data
Virtual streaming interface
Virtual hardware
Event record
Performance analysis
Source code
Demo
Resources
Questions
Strategic Partners