Object segmentation and reconstruction via an oscillatory neural network: interaction among learning, memory, topological organization and γ-band synchronization

摘要: Synchronization of neuronal activity in the γ-band has been shown to play an important role higher cognitive functions, by grouping together necessary information different cortical areas achieve a coherent perception. In present work, we used neural network Wilson-Cowan oscillators analyze problem binding and segmentation high-level objects. Binding is achieved implementing similarity prior knowledge Gestalt rules. Similarity law realized via topological maps within network. Prior originates means Hebbian rule synaptic change; objects are memorized with strengths. Segmentation global inhibitor which allows desynchronisation among multiple avoiding interference. Simulation results performed 40 x grid, using three simultaneous input objects, show that able recognize segment several conditions (different degrees incompleteness or distortion patterns), exhibiting reconstruction performances strength object memory. The presented model represents integrated approach for investigating relationships learning, memory, organization synchronization.