Using Spotted Hyena Optimizer for Training
Feedforward Neural Networks
Jie Li
1
, Qifang Luo
1,2(&)
, Ling Liao
1
, and Yongquan Zhou
1,2
1
College of Information Science and Engineering,
Guangxi University for Nationalities, Nanning 530006, China
l.qf@163.com
2
Guangxi High School Key Laboratory of Complex System and Computational
Intelligence, Nanning 530006, China
Abstract. Spotted hyena optimizer (SHO) is a novel heuristic optimization
algorithm based on the behavior of spotted hyena and their collaborative behavior
in nature. In this paper, we design a spotted hyena optimizer for feedforward
neural networks (FNNs). Training feedforward neural networks is regard as a
challenging task, because it is easy to fall into local optima. Our objective is to
apply heuristic optimization algorithm design to tackle these problems better than
the mathematical and deterministic methods, in order to confirm that SHO algo-
rithm training FNN is more effective. a classification datasets about Heart is
applies to benchmark the performance of the proposed method. The more basic
SHO is compared to other acclaimed state-of-the-art optimization algorithm, the
results show that the proposed algorithm can provide better results.
Keywords: Spotted hyena optimizer
Feedforward neural networks
Classification datasets
Function-approximation
1 Introduction
One of the most important inventions in the field of soft computing is neural networks.
(NN), It is inspired by the neurons of human brain. The basic concept of NN is first to
establish a mathematical model by McCulloch and Pitts [1]. In feedforward neural
network (FNN) [2], this is the focus of the problem study, the connection weights
between different layers of neural network and the biases of neural network layers are
the most important parameters.
There are more different types of heuristics that have emerged on training FNNs,
such as Montana and Davis was first applies GA to improve learning of NNs [3].
The PSO algorithm was also employed as the trainer for FNN in many studies [4].
Some other heuristic-based learning algorithms in the literature are as follows: DE
based trainer [5], gravitational search algorithm (GSA)-based trainer [6] and grey wolf
optimizer (GWO) [7] so on. Despite the many advantages of metaheuristic algorithms,
the problem of falling into the local optimum still exists.
Spotted hyena optimizer (SHO) [8] is a new metaheuristic optimization method, SHO
is inspired from social hierarchy and hunting behavior of spotted hyena. Select SHO
training feedforward neural network is derived from the proposed algorithm has a strong
© Springer International Publishing AG, part of Springer Nature 2018
D.-S. Huang et al. (Eds.): ICIC 2018, LNAI 10956, pp. 828–833, 2018.
https://doi.org/10.1007/978-3-319-95957-3_88