Jwa Yeon-Jae, Di Guglielmo Giuseppe, Arnold Lukas, Carloni Luca, Karagiorgi Georgia
Columbia University, New York, NY, United States.
Front Artif Intell. 2022 May 18;5:855184. doi: 10.3389/frai.2022.855184. eCollection 2022.
We present a custom implementation of a 2D Convolutional Neural Network (CNN) as a viable application for real-time data selection in high-resolution and high-rate particle imaging detectors, making use of hardware acceleration in high-end Field Programmable Gate Arrays (FPGAs). To meet FPGA resource constraints, a two-layer CNN is optimized for accuracy and latency with KerasTuner, and network is further used to minimize the computing resource utilization of the network. We use "High Level Synthesis for Machine Learning" () tools to test CNN deployment on a Xilinx UltraScale+ FPGA, which is an FPGA technology proposed for use in the front-end readout system of the future Deep Underground Neutrino Experiment (DUNE) particle detector. We evaluate network accuracy and estimate latency and hardware resource usage, and comment on the feasibility of applying CNNs for real-time data selection within the currently planned DUNE data acquisition system. This represents the first-ever exploration of employing 2D CNNs on FPGAs for DUNE.
我们展示了一种二维卷积神经网络（CNN）的定制实现，它是高分辨率和高速粒子成像探测器中实时数据选择的可行应用，利用高端现场可编程门阵列（FPGA）中的硬件加速。为满足FPGA资源限制，使用KerasTuner对两层CNN进行了精度和延迟优化，并进一步用于最小化网络的计算资源利用率。我们使用“机器学习高级综合”（HLS）工具在赛灵思UltraScale + FPGA上测试CNN部署，该FPGA技术被提议用于未来深层地下中微子实验（DUNE）粒子探测器的前端读出系统。我们评估网络精度，估计延迟和硬件资源使用情况，并评论在当前计划的DUNE数据采集系统中应用CNN进行实时数据选择的可行性。这代表了首次在FPGA上为DUNE采用二维CNN的探索。
