Object selection by oscillatory correlation

One of the classical topics in neural networks is winner-take-all (WTA), which has been widely used in unsupervised (competitive) learning, cortical processing, and attentional control. With global connectivity WTA networks, however, do not encode spatial relations in the input, and thus cannot support sensory and perceptual processing where spatial relationships are important. We propose a new architecture that maintains spatial relations. This selection network builds on oscillatory dynamics and slow inhibition In an input scene with many objects, the network selects the largest object. The system can be easily adjusted to select several largest objects, which then alternate in time. The network is applied successfully to select the most salient objects in real images