Overview
Machine learning (ML) algorithms have become essential and ubiquitous components of physics experiments, especially in time-critical and resource-constrained applications in trigger, data acquisition, or on-detector (“edge”) systems. Often, it is necessary to deploy these algorithms in experiment-specific targeted platforms, including field-programmable gate arrays (FPGAs).
In this module, students will learn how to train an ML algorithm for an experimental physics task and then consider its latency, throughput, and resource usage.
Students will apply quantization-aware training and parameter pruning to compress the model, making it faster and more efficient, while maintaining an acceptable accuracy. Finally, students will use the hls4ml Python library to deploy the algorithm on a PYNQ-Z2 FPGA development board, as shown above.
Lab Kit Inventory
For each of the 4 lab stations, we will require the following hardware and software
