2 Comments

Очень верно! Нужно смоделировать формирование нейронной сети, управляемое входным потоком воздействий. То есть, руководствуясь небольшим набором простых правил, вроде - связь "прорастает" к возбужденному нейрону-соседу, сформировавшаяся связь запрещает "прорастание" дублирующей связи, если оба нейрона (источник и получатель) при формировании связи активны, то связь получается возбуждающей, иначе - тормозящей и т.п. Жаль, у меня не хватило упорства построить такую (демонстрационную) модель.

Expand full comment

Mykola, programming systems could be used to solve problems where information explosion is not the problem.

Attempting to increase the number of resources allocated to solve a task and improving a set of governing algorithms could help to some extent but not in general.

Only transition to the development of artificial subjective systems could open the opportunity to have artificial systems capable to deal with an unconstrained world.

Improvements cannot help to solve the Perpetual Motion design problem.

In addition:

Brains do not have any business in the real world and cannot work with information about events in it because did not have direct access to it and senses do not obtain information about causes leading to changes in its state.

Previously I explained what properties of artificial systems made them universal. One more thing about them. It is impossible to have an adequate model of all the known world within any system. The natural and artificial systems did not exclude.

However, subjective systems do not possess that problem.

I could explain why to my partners if it will be possible to find them.

Expand full comment