Optimization with GMDH

Polynomial Neural Networks are one of the most sought after techniques for development of numerical,non-linear and cognative models. Group Method of Data Handling or GMDH is an example algorithm which follows the architecture.

In the basic architecture followed in neural networks the output is estimated by the weighted sum of the inputs where transformation from input to output function is conducted by some linear functions. "The processing function of the neurons is quite simple; it is the configuration of the network itself that requires much work to design and adjust to the training data."(Link to the full story)

Frank Rosenblatt was responsible for identification of the key weakness of "neuro-computing as the lack of means for effectively selecting structure and weights of the hidden layer(s) of the perceptron". In 1968,after seven years from the time Rosenblatt's weakness identification, when back-propagation technique was not known yet, a method "called Group Method of Data Handling (GMDH) was developed by an Ukranian scientist Aleksey Ivakhnenko who was working at that time on a better prediction of fish population in rivers."

He made the neuron "a more complex unit featuring a polynomial transfer function" but the interconnections between layers of neurons were simplified, and an self adaptive selection algorithm for structure design and weight adjustment was developed.(Link to the full story)

There are many papers which highlights the application of GMDH in optimization but still their is much scope for applying GMDH for optimization.

Find some new papers of 'application of GMDH in optimization' in UO.

Thanks,
Admin and Editor of
Utilize Optimally : Learn about Optimization by Soft-computation
Water based Renewable Energies
Call for Papers Special Issue
Subscribe to the newsletter/Contact me/Advertise/Call for Paper/Invitation to act as a contributor/My Author Profile/My Kudos Profile