To cite this Article De, Sourav, Bhattacharyya, Siddhartha and Dutta, Paramartha(2010) 'Efficient grey-level image segmentation using an optimised MUSIG (OptiMUSIG) activation function', International Journal of Parallel, Emergent and Distributed Systems,, First published on: 22 February 2010 (iFirst) To link to this Article: DOI: 10.1080/17445760903546618 URL: http://dx.doi.org/10.1080/1744576...

The Liouville theorem states that bounded holomorphic complex functions are necessarily constant. Holomorphic functions fulfill the socalled Cauchy-Riemann (CR) conditions. The CR conditions mean that a complex z-derivative is independent of the direction. Holomorphic functions are ideal for activation functions of complex neural networks, but the Liouville theorem makes them useless. Yet recen...

The activation can be rewritten as follows a = vx+ b where v = [w1, . . . , wD] and b = wD+1 , and is an inner product between a weight vector v and the input x plus a bias b. For different inputs x of the same magnitude1, the activation is maximum when x is parallel to v, and the latter can be viewed as a pattern or template to which x is compared. The bias b then raises or lowers the activati...

j=1 aijgj(uj(t)) + Ii (i = 1, · · · , n) (1, 1) where gj(x) are activation functions, di, aij are constants and Ii are constant inputs. In practice, however, the interconnections contain asynchronous terms in general, and the interconnection weights aij, bij , self-inhibition di and inputs Ii should depend on time, often periodically. Therefore, we need to discuss the following dynamical system...

In this paper we have defined a novel activation function called the multi-level signum for the real and complex valued associative memories. The major motivation of such a function is to increase the number of patterns that can be stored in a memory without increasing the number of neurons. The state of such a network can be described as one of the points that lie on a complex bounded lattice....

In this paper, a new algorithm for function approximation is proposed to obtain better generalization performance and faster convergent rate. The new algorithm incorporates the architectural constraints from a priori information of the function approximation problem into Extreme Learning Machine. On one hand, according to Taylor theorem, the activation functions of the hidden neurons in this al...

This work describes how to train the activation function in neuro-wavelet para-metric modeling. Training activation function signiicantly improves performance in a number of modeling, classii-cation and forecasting problems. Three diierent case studies from as many different application domains are considered and their performance compared with non-trained activation functions.

introduction: photon attenuation in tissues is the primary physical degrading factor limiting both visual qualitative interpretation and quantitative analysis capabilities of reconstructed single photon emission computed tomography (spect) images. the aim of present study was to investigate the effect of attenuation correction on the detection of activation foci following statistical analysis w...

cellulose wastes of a wood and paper factory were used to produce activated carbon. several chemical agents under various conditions were used for production of activated carbon and hence their adsorption properties have been evaluated. in addition the effect of process parameters such as raw material, chemical agent concentration, impregnation ratio, carbonization temperature, carbonization ti...