Hardware-friendly Higher-Order Neural Network Training using Distributed Evolutionary Algorithms
Fecha
2010Materia
Resumen
In this paper, we study the class of Higher-Order Neural Networks and especially the Pi-Sigma Networks. The performance of Pi-Sigma Networks is evaluated through several well known Neural Network Training benchmarks. In the experiments reported here, Distributed Evolutionary Algorithms are implemented for Pi-Sigma neural networks training. More specifically the distributed versions of the Differential Evolution and the Particle Swarm Optimization algorithms have been employed. To this end, each processor is assigned a subpopulation of potential solutions. The subpopulations are independently evolved in parallel and occasional migration is employed to allow cooperation between them. The proposed approach is applied to train Pi-Sigma Networks using threshold activation functions. Moreover, the weights and biases were confined to a narrow band of integers, constrained in the range [-32; 32]. Thus, the trained Pi-Sigma neural networks can be represented by using 6 bits. Such networks are better suited than the real weight ones for hardware implementation and to some extend are immune to low amplitude noise that possibly contaminates the training data. Experimental results suggest that the proposed training process is fast, stable and reliable and the distributed trained Pi-Sigma Networks exhibited good generalization capabilities. (C) 2009 Elsevier B.V. All rights reserved.
Colecciones
Ítems relacionados
Mostrando ítems relacionados por Título, autor o materia.
-
CONTENT project: Considerations towards a cloud-based internetworking paradigm
Katsalis, K.; Korakis, T.; Landi, G.; Bernini, G.; Rofoee, B. R.; Peng, S.; Anastasopoulos, M.; Tzanakaki, A.; Christofi, D.; Georgiades, M.; Larsen, R.; Riera, J. F.; Escalona, E.; Garcia-Espin, J. A. (2013)Although cloud computing and the Software Defined Network (SDN) framework are fundamentally changing the way we think about network services, multi-domain and multitechnology problems are not sufficiently investigated. ... -
Backbone formation in military multi-layer ad hoc networks using complex network concepts
Papakostas D., Basaras P., Katsaros D., Tassiulas L. (2016)Modern battlefields are characterized by increasing deployment of ad hoc communications among allied entities. These networks can be seen as a complex multi-layer ad hoc network, where each layer may be an independently ... -
Query sensitive storage for wireless sensor networks
Papadimitriou, A.; Katsaros, D.; Manolopoulos, Y. (2009)Storage management in wireless sensor networks is an area that has started to attract significant attention, and several methods have been proposed, such as Local Storage (LS), Data-Centric Storage (DCS) and more recently ...