350 rub
Journal Neurocomputers №4 for 2015 г.
Article in number:
In favor of Neurocomputing
Keywords:
neurocomputing
pseudoscientific
the least squares method
approximation
the threshold functions basis.
Authors:
I.S. Bryantcev - master\'s degree, Moscow City Pedagogical University (MСPU). E-mail: kolbasa-90@list.ru
A.V. Savelyev - Senior Research Scientist, Director of the Patent agency «©Uniquely honest patenting» (Moscow). E-mail: gmkristo@rambler.ru
Abstract:
The reason that NC-technology often become the of pseudoscience accusations is the prefix «neuro-» only. Usually under the «neural network technologies» and «Neurocomputing» implies an approximation of the empirical data by least squares in the basis of threshold functions and the mathematics methods of Newton, Euler, at no later than 16-18 centuries. They are, it is not adapted to the existing computer equipment, and they are designed for manual account, or using a slide rule, so it is ignored different principles of information processing by computers. The prefix «neuro-» used exclusively branded promotional purposes, exploiting usually phonetic harmony and current fashion only, and nothing to neurons has not. Another approach - a fantasy are unrelated to computational mathematics. Unsolvable problems, on the one hand, and the appeal of drug formulation creates endless motive, but fruitless solving attempts. Because the main thing is solving process that and it can send the activities energy for using with any usually selfish reasons. It is noteworthy that in this and in another case to the brain and «neuro» is irrelevant. Rather, the second group can bring to real problems still somehow (such as Fermat\'s theorem or the theory of divisors). The first group more real and allows to receive real, albeit partial results. But it is doomed to wander in constant hard within a very narrowly circumscribed mathematical paradigm - approximation empiric data. It is\'t not approaching the neurons, nor to the «neural networks», nor to the brain and do not have to do with them. And above all - the mythical «formal neuron» McCulloch - Pitts does not exist in reality, but only in the imagination of man [1].In this sense calling him «neuron» is no less fantastic than the «transmigration of the soul into the computer». It is disguised more that purely linguistically and not so obviously hurts the eyes.
In this case, if we want to qualify for the term «Neurocomputing» we should be interested in the subject of research necessary or at least a little (if the research somehow relates to the object) and we should be approaching the real biological nervous system, or to build up from the term «neuro» and engage their work. Otherwise, we will continue to receive «neuro-oil pumps», «neuro-gas turbine engines» «neuro-traffic light», «neuro-learning process», and other neuro-nonsense which were discredited normal Neurocomputing.
This work was financially supported by RHSF in grant № 15-03-00519а «Post-non-classic paradigm of artificial intellect».
Pages: 24-25
References
- Savelev A.V. Nejjrosociometodologija: problemy dialoga mezhdu nejjrobiologiejj i nejjromodelirovaniem // Nejjrokompjutery: razrabotka, primenenie. 2011. № 1. S. 47-63.
- Kolushov V.V., Savelev A.V. Primenenie 4D-vizualizacionnogo konstruirovanija sverkhslozhnykh sistem v biologii i medicine // Biomedicinskaja radioehlektronika. 2014. № 4. S. 35-37.