Hybrid extreme learning machine approach for homogeneous neural networks

Christou, Vasileios; Tsipouras, Markos G.; Γιαννακέας, Νικόλαος; Tzallas, Alexandros T.

doi:10.1016/j.neucom.2018.05.064

Cited by 10 publications

(13 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this section, a structured view of the custom neurons, that are utilized in this research, is presented. It divides neurons into three subcomponents (the last two form the neuron's soma S) according to the structured composite model (C-Model) proposed by Christou et al [9]. These neuron subcomponents are the dendrite (D), the activation (S a ) function and the activation-output (S ao ) function.…”

Section: He-hyelm Structural Unitsmentioning

confidence: 99%

“…The weighted input vectors form the neuron's dendrite which is summed by the activation function and passed as input to the activation-output function. This study used two different types of dendrites, the linear (D l ) dendrite and a variation of the cubic (D c ) dendrite called multi-cube (D mc ) dendrite [9] which are going to be analysed in Sections 3.2.1 and 3.2.2. The distinction between different types of dendrites is done utilizing the l, c and mc superscripts which define the linear, cubic and multi-cubic types accordingly.…”

Section: Varying the Dendritesmentioning

confidence: 99%

“…This bit string takes only the values -1 and 1. Finally, with n we define the number of inputs [19,9].…”

Section: Cubic Dendritesmentioning

confidence: 99%

“…This approach has shown to be effective for both regression and classification problems although it may require more nodes than networks trained with gradient-based methods like back-propagation [24] due to the randomization of the hidden layer weights and thresholds [72,69]. Furthermore, the ELM algorithm is unable to create SLNNs containing different combinations of neuron types 1 in the hidden layer (these network types will 1 The term different combinations of neuron types refers to neuron types with structural units like the structured composite model (C-Model) proposed by Christou et al [9] which is utilized in the proposed algorithm. This model divides the neuron into three subunits (subcomponents) named dendrite (D), activation (S a ) function, activation-output (S ao ) function and is able to create a custom neuron by varying any of the subunits from each category.…”

Section: Introductionmentioning

confidence: 99%

“…The Ho-HyELM approach proposed by Christou et al [9] can create networks with different neuron types in the hidden layer but is restricted in the creation of homogeneous hidden layers.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Hybrid extreme learning machine approach for heterogeneous neural networks

et al. 2019

Self Cite

View full text Add to dashboard Cite

In this paper, a hybrid learning approach, which combines the extreme learning machine (ELM) with a genetic algorithm (GA), is proposed. The utilization of this hybrid algorithm enables the creation of heterogeneous single layer neural networks (SLNNs) with better generalization ability than traditional ELM in terms of lower mean square error (MSE) for regression problems or higher accuracy for classification problems. The architecture of this method is not limited to traditional linear neurons, where each input participates equally to the neuron's activation, but is extended to support higher order neurons which affect the network's generalization ability. Initially, the proposed heterogeneous hybrid extreme learning machine (He-HyELM) algorithm creates a number of custom created neurons with different structure, which are used for the creation of homogeneous SLNNs. These networks are trained with ELM and an application specific GA evolves them into heterogeneous networks according to a fitness criterion utilizing the uniform crossover operator for the recombination process. After the completion of the evolution process, the network with the best fitness is selected as the most optimal.

show abstract

Section: He-hyelm Structural Unitsmentioning

confidence: 99%

Section: Varying the Dendritesmentioning

confidence: 99%

“…This bit string takes only the values -1 and 1. Finally, with n we define the number of inputs [19,9].…”

Section: Cubic Dendritesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

“…The Ho-HyELM approach proposed by Christou et al [9] can create networks with different neuron types in the hidden layer but is restricted in the creation of homogeneous hidden layers.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations