Explore an open-source software-hardware co-design workflow called hls4ml, designed to empower scientific low-power machine learning devices. Learn about the essential features of this workflow, including network optimization techniques such as pruning and quantization-aware training. Discover how hls4ml supports domain scientists by interpreting and translating machine learning algorithms for implementation in FPGAs and ASICs. Gain insights into the expanded capabilities of hls4ml, including new Python APIs, quantization-aware pruning, end-to-end FPGA workflows, long pipeline kernels for low power, and new device backends with an ASIC workflow. Understand how these advancements in hls4ml aim to provide accessible, efficient, and powerful tools for machine-learning-accelerated scientific discovery across various application domains.
hls4ml: An Open-Source Co-Design Workflow for Scientific Low-Power Machine Learning Devices
Overview
Syllabus
Introduction
Motivation
Features
Questions
Sponsors
Taught by
tinyML
Related Courses
-
tinyMLPerf: Benchmarking Ultra-low Power Machine Learning Systems
-
AI at the Edge - Enabling Vision for Low-Power Devices
-
Power-of-Two Quantization for Low Bitwidth and Hardware Compliant Neural Networks
-
Co-design Hardware and Algorithm for Vector Search
-
The Akida Neural Processor - Low Power CNN Inference and Learning
-
TinyML Talks Germany - AutoFlow - An Open Source Framework to Automatically Implement Neural
