The unsupervised learning of feature extraction in high-dimesional patterns is a central problem for the neural network approach. Feature extraction is a procedure which maps original patterns into the feature (or factor) space of reduced dimension. In this paper we demonstrate that Hebbian learning in Hopfield-like neural network is a natural procedure for unsupervised learning of feature extraction. Due to this learning, factors become the attractors of network dynamics, hence they can be revealed by the random search. The neurodynamics is analysed by Single-Step approximation which is known [8] to be rather accurate for sparsely encoded Hopfield-network. Thus, the analysis is restricted by the case of sparsely encoded factors. The accuracy of Single-Step approximation is confirmed by Computer simulations.
Movement learning results from synaptic plasticities in various sites of the brain. Three sites háve been particularly studied: the cortico-cortical synapses in the cerebral cortex, the parallel fiber-Purkinje cell synapses in the cerebellar cortex and the cerebello-thalamo-cortical pathway at the level of the thalamocortical synapses. We intended to understand how these three adaptive processes cooperate for optimal performance during the arm reaching movement, and how the cerebellar learning is supervised. A neural network model was developed on the basis of two main prerequisites: the columnar organization of the cerebral cortex and the Marr-Albus-Ito theory of cerebellar learning. The synaptic plasticities observed on these sites were incorporated in the model as differential equations. The analytical resolution of the set of rules showed two main results. First, the adaptive processes taking place in different sites do not interfere but complement each other during the learning of the arm reaching movement. Secondly, any linear combination of the cerebral motor commands may generate olivary signals able to supervise the cerebellar learning process.
A sparsely encoded Willshaw-like attractor neural network based on the binary Hebbian synapses is investigated analytically and by Computer simulations. A special inhibition mechanism which supports a constant number of active neurons at each time step is used. The informationg capacity and the size of attraction basins are evaluated for the Single-Step and the Gibson-Robinson approximations, as well as for experimental results.
Motor recovery in post-stroke and post-traumatic patients using exoskeleton controlled by the brain-computer interface (BCI) is a new and promising rehabilitation procedure. Its development is a multidisciplinary research which requires, the teamwork of experts in neurology, neurophysiology, physics, mathematics, biomechanics and robotics. Some aspects of all these fields of study concerning the development of this rehabilitation procedure are described in the paper. The description includes the principles and physiological prerequisites of BCI based on motor imagery, biologically adequate principles of exoskeleton design and control and the results of clinical application.