Epsilon greedy strategy for hyper parameters tuning of a neural network equalizer

Quyet Nguyen, Noel Teku, Tamal Bose

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In wireless communications, equalization can be used to remove channel impairments from transmissions. Neural networks (NNs) have proven to be an effective technique against conventional equalizers (i.e. decision-feedback, zero-forcing, etc.). High Frequency (HF) channels require high-performance equalizers to overcome Doppler shifts and large delay spreads. When using a NN equalizer, tuning its structure (i.e. activation function, optimizer, etc...) can be time-consuming. This work proposes using an annealing epsilon greedy algorithm, a reinforcement learning technique, to tune the attributes of a neural network equalizer. Reinforcement learning has been used to tune NNs in different applications, but to the best of our knowledge, it has not been done for NN equalization. The objective of this work is to analyze if using reinforcement learning can improve the performance of a NN equalizer.

Original languageEnglish (US)
Title of host publicationISPA 2021 - 12th International Symposium on Image and Signal Processing and Analysis
EditorsTomislav Petkovic, Davor Petrinovic, Sven Loncaric
PublisherIEEE Computer Society
Pages209-212
Number of pages4
ISBN (Electronic)9781665426398
DOIs
StatePublished - Sep 13 2021
Event12th International Symposium on Image and Signal Processing and Analysis, ISPA 2021 - Virtual, Zagreb, Croatia
Duration: Sep 13 2021Sep 15 2021

Publication series

NameInternational Symposium on Image and Signal Processing and Analysis, ISPA
Volume2021-September
ISSN (Print)1845-5921
ISSN (Electronic)1849-2266

Conference

Conference12th International Symposium on Image and Signal Processing and Analysis, ISPA 2021
Country/TerritoryCroatia
CityVirtual, Zagreb
Period9/13/219/15/21

Keywords

  • Epsilon greedy algorithm
  • High frequency (HF) channel
  • Neural network equalizer
  • Reinforcement learning

ASJC Scopus subject areas

  • Computational Theory and Mathematics
  • Computer Graphics and Computer-Aided Design
  • Computer Vision and Pattern Recognition
  • Signal Processing

Fingerprint

Dive into the research topics of 'Epsilon greedy strategy for hyper parameters tuning of a neural network equalizer'. Together they form a unique fingerprint.

Cite this