Heuristic hyperparameter optimization for multilayer perceptron with one hidden layer

Łukasz Neumann; Robert M. Nowak

doi:10.1117/12.2501569

1 October 2018 Heuristic hyperparameter optimization for multilayer perceptron with one hidden layer

Łukasz Neumann, Robert M. Nowak

Proceedings Volume 10808, Photonics Applications in Astronomy, Communications, Industry, and High-Energy Physics Experiments 2018; 108082A (2018) https://doi.org/10.1117/12.2501569
Event: Photonics Applications in Astronomy, Communications, Industry, and High-Energy Physics Experiments 2018, 2018, Wilga, Poland

Abstract

One of the crucial steps of preparing a neural network model is the process of tuning its hyperparameters. This process can be time-consuming and hard to be done properly by hand. Tuned hyperparameters allow to obtain high accuracy of classification as well as fast training. In this paper we explore the usage of selected heuristic algorithms based on evolutionary approach: Covariance Matrix Adaptation Evolution Strategy (CMAES), Differential Evolution Strategy (DES) and jSO for the hyperparameter tuning task. Results of Multilayer Perceptron’s (MLP) hyperparameter optimization for a real-life dataset are presented. An improvement in models’ performance is observed through the usage of presented approach.

Citation Download Citation

Łukasz Neumann and Robert M. Nowak "Heuristic hyperparameter optimization for multilayer perceptron with one hidden layer", Proc. SPIE 10808, Photonics Applications in Astronomy, Communications, Industry, and High-Energy Physics Experiments 2018, 108082A (1 October 2018); https://doi.org/10.1117/12.2501569

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available