Reduced Precision Strategies for Deep Learning: A High Energy Physics Generative Adversarial Network Use Case

Florian Rehm, Florian Rehm, Sofia Vallecorsa, Vikram Saletore, Hans Pabst, Adel Chaibi, Valeriu Codreanu, Kerstin Borras, Kerstin Borras, Dirk Krücker

Abstract

Deep learning is finding its way into high energy physics by replacing traditional Monte Carlo simulations. However, deep learning still requires an excessive amount of computational resources. A promising approach to make deep learning more efficient is to quantize the parameters of the neural networks to reduced precision. Reduced precision computing is extensively used in modern deep learning and results to lower execution inference time, smaller memory footprint and less memory bandwidth. In this paper we analyse the effects of low precision inference on a complex deep generative adversarial network model. The use case which we are addressing is calorimeter detector simulations of subatomic particle interactions in accelerator based high energy physics. We employ the novel Intel low precision optimization tool (iLoT) for quantization and compare the results to the quantized model from TensorFlow Lite. In the performance benchmark we gain a speed-up of 1.73x on Intel hardware for the quantized iLoT model compared to the initial, not quantized, model. With different physics-inspired self-developed metrics, we validate that the quantized iLoT model shows a lower loss of physical accuracy in comparison to the TensorFlow Lite model.

Download


Paper Citation


in Harvard Style

Rehm F., Vallecorsa S., Saletore V., Pabst H., Chaibi A., Codreanu V., Borras K. and Krücker D. (2021). Reduced Precision Strategies for Deep Learning: A High Energy Physics Generative Adversarial Network Use Case.In Proceedings of the 10th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM, ISBN 978-989-758-486-2, pages 251-258. DOI: 10.5220/0010245002510258


in Bibtex Style

@conference{icpram21,
author={Florian Rehm and Sofia Vallecorsa and Vikram Saletore and Hans Pabst and Adel Chaibi and Valeriu Codreanu and Kerstin Borras and Dirk Krücker},
title={Reduced Precision Strategies for Deep Learning: A High Energy Physics Generative Adversarial Network Use Case},
booktitle={Proceedings of the 10th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM,},
year={2021},
pages={251-258},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0010245002510258},
isbn={978-989-758-486-2},
}


in EndNote Style

TY - CONF

JO - Proceedings of the 10th International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM,
TI - Reduced Precision Strategies for Deep Learning: A High Energy Physics Generative Adversarial Network Use Case
SN - 978-989-758-486-2
AU - Rehm F.
AU - Vallecorsa S.
AU - Saletore V.
AU - Pabst H.
AU - Chaibi A.
AU - Codreanu V.
AU - Borras K.
AU - Krücker D.
PY - 2021
SP - 251
EP - 258
DO - 10.5220/0010245002510258