Informatica

The more realistic neural soma and synaptic nonlinear relations and an alternative mean field theory (MFT) approach relevant for strongly interconnected systems as a cortical matter are considered. The general procedure of averaging the quenched random states in the fully-connected networks for MFT, as usually, is based on the Boltzmann Machine learning. But this approach requires an unrealistically large number of samples to provide a reliable performance. We suppose an alternative MFT with deterministic features instead of stochastic nature of searching a solution a set of large number equations. Of course, this alternative theory will not be strictly valid for infinite number of elements. Another property of generalization is an inclusion of the additional member in the effective Hamiltonian allowing to improve the stochastic hill-climbing search of the solution not dropping into local minima of the energy function. Especially, we pay attention to increasing of neural networks retrieval capability transforming the replica-symmetry model by including of different nonlinear elements. Some results of numerical modeling as well as the wide discussion of neural systems storage capacity are presented.

The nonlinearities play a crucial role in the brain processes. They take place in neuronal system elements: synapses, dendrite membranes, soma of neurons, axons. It is established that the soma nonlinearity, which is of sigmoidal shape, is not so strong as compared with the electric current-voltage relation of a dendrite membrane. The relation is N-shaped with two stable and one unstable points. In dynamics, this leads to the appearance of a switch wave or formation of some logic functions. We present some artificial logic circuits based on an electrical analogy of dendritic membrane characteristics in static and dynamic cases. The nonlinear cable theory and the numerical simulation were used. Basing on the logic circuit construction proposed, we suppose that the dendritic membrane processes are able not only to gather and transfer information but also to transform and classify knowledge.

This discussion brings together three different perspectives, – from cognitive science, ergonomics and learning – that are crucial to consider in the process of designing hypertext as a constructivist environment for learning. The critical implications from all the perspectives are illustrated in several screen shots from one concrete hypertext application for learning of Pascal programming.