Hyperparameter optimization, quantum-assisted model performance prediction, and benchmarking of AI-based High Energy Physics workloads using HPC

TL;DR

This paper explores using quantum-assisted performance prediction to optimize hyperparameters for AI models in high energy physics workloads on HPC, demonstrating potential integration of quantum and classical methods and benchmarking hardware accelerators.

Contribution

It introduces a quantum-enhanced performance predictor for HPO, proposes solutions to quantum system limitations, and develops a benchmark for hardware comparison in AI workloads.

Findings

Quantum-assisted predictor achieves results comparable to classical methods.

Proposed method improves stability of quantum solutions.

Benchmark enables hardware assessment for deep learning in physics.

Abstract

Training and Hyperparameter Optimization (HPO) of deep learning-based AI models are often compute resource intensive and calls for the use of large-scale distributed resources as well as scalable and resource efficient hyperparameter search algorithms. This work studies the potential of using model performance prediction to aid the HPO process carried out on High Performance Computing systems. In addition, a quantum annealer is used to train the performance predictor and a method is proposed to overcome some of the problems derived from the current limitations in quantum systems as well as to increase the stability of solutions. This allows for achieving results on a quantum machine comparable to those obtained on a classical machine, showing how quantum computers could be integrated within classical machine learning tuning pipelines. Furthermore, results are presented from the development of a containerized benchmark based on an AI-model for collision event reconstruction that allows us to compare and assess the suitability of different hardware accelerators for training deep neural networks.