Autor: |
Timea Bezdan, Nikola Vukobrat, Ivana Strumberger, Nebojsa Bacanin, Miodrag Zivkovic |
Rok vydání: |
2021 |
Předmět: |
|
Zdroj: |
Intelligent and Fuzzy Techniques for Emerging Conditions and Digital Transformation ISBN: 9783030855765 |
Popis: |
The learning process is one of the most difficult problems in artificial neural networks. This process goal is to find the appropriate values for connection weights and biases and has a direct influence on the neural network classification and prediction accuracy. Since the search space is huge, traditional optimization techniques are not suitable as they are prone to slow convergence and getting trapped in the local optima. In this paper, an enhanced harris hawks optimization algorithm is proposed to address the task of neural networks training. Conducted experiments include 2 well-known classification benchmark datasets to evaluate the performance of the proposed method. The obtained results indicate that the devised algorithm has promising performance, as that it is able to achieve better overall results than other state-of-the-art metaheuristics that were taken into account in comparative analysis, in terms of classification accuracy and converging speed. |
Databáze: |
OpenAIRE |
Externí odkaz: |
|