Abstract

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. Because of global connectivity, WTA networks, however, do not encode spatial relations in the input, and thus cannot support sensory and perceptual processing where spatial relations are important. We propose a new architecture that maintains spatial relations between input features. This selection network builds on LEGION (Locally Excitatory Globally Inhibitory Oscillator Networks) dynamics and slow inhibition. In an input scene with many objects (patterns), the network selects the largest object. This system can be easily adjusted to select several largest objects, which then alternate in time. We further show that a two-stage selection network gains efficiency by combining selection with parallel removal of noisy regions. The network is applied to select the most salient object in real images. As a s...

Citations

399	An introduction to computing with neural nets - Lippmann - 1987
396	Shifts in selective visual attention: Towards the under-laying neural circuitry, Human Neurobiology 4 - Koch, Ullman
342	Adaptive pattern classification and universal recoding - Grossberg - 1976
339	J: Neural mechanisms of selective visual attention. Annu Rev Neurosci - Desimone, Duncan - 1995
274	Corticonics: neural circuits of the cerebral cortex - Abeles, ed - 1991
251	Feature discovery by competitive learning - Rumelhart, Zipser - 1985
114	Synchronization of cortical activity and its putative role in information processing and learning - Singer - 1993
102	The correlation theory of brain function - Malsburg - 1981
88	Global competition and local cooperation in a network of neural oscillators - Terman, Wang - 1995
71	Alternating and synchronous rhythms in reciprocally inhibitory model neurons, Neural Comput - Wang, Rinzel - 1992
61	Why are small and large numbers enumerated differently? A limited-capacity preattentive stage in vision - Trick, Pylyshyn - 1994
60	Rethinking perceptual organization: The role of uniform connectedness - Palmer, Rock - 1994
59	Locally excitatory globally inhibitory oscillator networks - Wang, Terman - 1995
57	A model for visual shape recognition - Milner - 1974
39	Emergent synchrony in locally coupled neural oscillators - Wang - 1995
38	Competition and cooperation in neural nets - Amari, Arbib - 1977
30	Dialogues on Perception - Julesz - 1995
28	An oscillation-based model for the neuronal basis of attention. Vision Res 33:2789–2802 - Niebur, Koch, et al. - 1993
18	Perceptual organization and attention - Mack, Tang, et al. - 1992
16	Complex dynamics in winner-take-all neural nets with slow inhibition, Neural Networks - Ermentrout
16	Towards a computational model of visual attention - Tsotsos - 1995
14	Fast numerical integration of relaxation oscillator networks based on singular limit solutions - Linsay, Wang - 1998
10	Subharmonic coordination in networks of neurons with slow conductances - LoFaro, Kopell, et al. - 1994
9	Shifts in selective visual attention - KOCH, ULLMAN - 1985
8	Features and objects: the fourteenth Barlett memorial lecture - Treisman
7	Partial synchronization in a network of neural oscillators - Terman, Lee - 1997
4	The simulation and modeling of distributed information processing in the frog visual system - Didday - 1970
2	Handbook of perception and action: Attention - Neumann, Sanders - 1996