Home // PESARO 2018, The Eighth International Conference on Performance, Safety and Robustness in Complex Systems and Applications // View article

Exploring Deep Neural Networks for Regression Analysis

Authors:
Florian Kästner
Benedikt Janßen
Frederik Kautz
Michael Hübner

Keywords: ANN; MLP; CNN; Reinforcement Learning

Abstract:
Designing artificial neural networks is a challenging task due to the vast design space. In this paper, we present our exploration on different types of deep neural networks and different shapes for a regression analysis task. The network types range from simple multi-layer perceptron networks to more complex convolutional and residual neural networks. Within the exploration, we analyzed the behavior of the different network shapes, when processing measurement data characteristic for mass spectrometers. Mass spectrometers are used to determine single substances within gaseous mixtures. By applying deep neural networks for the measurement data processing, the behavior of the measurement system can be approximated indirectly through the learning process. In addition, we evaluate the usage of reinforcement learning to design the neural network's architecture.

Pages: 1 to 6

Copyright: Copyright (c) IARIA, 2018

Publication date: April 22, 2018

Published in: conference

ISSN: 2308-3700

ISBN: 978-1-61208-628-6

Location: Athens, Greece

Dates: from April 22, 2018 to April 26, 2018