TY - GEN
T1 - A Novel Two-Level Clustering-Based Differential Evolution Algorithm for Training Neural Networks
AU - Mousavirad, Seyed Jalaleddin
AU - Oliva, Diego
AU - Schaefer, Gerald
AU - Moghadam, Mahshid Helali
AU - El-Abd, Mohammed
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.
PY - 2024/3/21
Y1 - 2024/3/21
N2 - Determining appropriate weights and biases for feed-forward neural networks is a critical task. Despite the prevalence of gradient-based methods for training, these approaches suffer from sensitivity to initial values and susceptibility to local optima. To address these challenges, we introduce a novel two-level clustering-based differential evolution approach, C2L-DE, to identify the initial seed for a gradient-based algorithm. In the initial phase, clustering is employed to detect some regions in the search space. Population updates are then executed based on the information available within each region. A new central point is proposed in the subsequent phase, leveraging cluster centres for incorporation into the population. Our C2L-DE algorithm is compared against several recent DE-based neural network training algorithms, and is shown to yield favourable performance.
AB - Determining appropriate weights and biases for feed-forward neural networks is a critical task. Despite the prevalence of gradient-based methods for training, these approaches suffer from sensitivity to initial values and susceptibility to local optima. To address these challenges, we introduce a novel two-level clustering-based differential evolution approach, C2L-DE, to identify the initial seed for a gradient-based algorithm. In the initial phase, clustering is employed to detect some regions in the search space. Population updates are then executed based on the information available within each region. A new central point is proposed in the subsequent phase, leveraging cluster centres for incorporation into the population. Our C2L-DE algorithm is compared against several recent DE-based neural network training algorithms, and is shown to yield favourable performance.
KW - clustering
KW - Differential evolution
KW - neural network training
KW - regularisation
UR - http://www.scopus.com/inward/record.url?scp=85189627585&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-56852-7_17
DO - 10.1007/978-3-031-56852-7_17
M3 - Conference contribution
AN - SCOPUS:85189627585
SN - 9783031568510
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 259
EP - 272
BT - Applications of Evolutionary Computation - 27th European Conference, EvoApplications 2024, Held as Part of EvoStar 2024, Proceedings
A2 - Smith, Stephen
A2 - Correia, João
A2 - Cintrano, Christian
PB - Springer Science and Business Media Deutschland GmbH
T2 - 27th European Conference on Applications of Evolutionary Computation, EvoApplications 2024
Y2 - 3 April 2024 through 5 April 2024
ER -