| layout | title | permalink | innav |
|---|---|---|---|
page |
About |
/about/ |
exclude |
Tiny Machine Learning (tinyML) is an emerging field of applied machine learning (ML) that focuses on running ML models on ultra-low-power embedded systems. The goal of this tutorial/workshop is to introduce people to the field of tinyML, showcase its unique applications and real-world use cases and dive into the underlying hardware and software that makes it tick. In this tutorial, we focus on efficient hardware and software deployment for tinyML systems. To that end, the first half of the tutorial dives into TensorFlow Lite Micro, an embedded machine learning framework for running ML models that are a few hundreds of KBs on tinyML hardware. The second half of the tutorial focuses on CFU Playground, a framework that an engineer, intern, or student can use to design and evaluate enhancements to an FPGA-based “soft” processor, specifically to increase the performance of machine learning (ML) tasks. This combination of hardware and software showcases the flexibility and strengths of open-source frameworks and hardware to rapidly explore the design space for developing model-specific accelerators for tinyML.
Contents
- TOC {:toc}
- What are some of the challenges and opportunities for designing tinyML hardware?
- How can we design and develop model-specific accelerators quickly on FPGAs?
- Get hands-on knowledge on how to build an ML accelerator using CFU playground
- Computer organization class students
- Computer architecture researcher students and practitioners
- Introduction to tinyML
- Introduction to TensorFlow Lite Micro
- Introduction to CFU Playground
- Hands-on tinyML on FPGAs