Temporal Album

E Vasilaki,JF Feng,H Buxton
DOI: https://doi.org/10.1109/tnn.2003.809641
2003-01-01
IEEE Transactions on Neural Networks
Abstract:Transient synchronization has been used as a mechanism of recognizing auditory patterns using integrate-and-fire neural networks. We first extend the mechanism to vision tasks and investigate the role of spike dependent learning. We show that such a temporal Hebbian learning rule significantly improves accuracy of detection. We demonstrate how multiple patterns can be identified by a single pattern selective neuron and how a temporal album can be constructed. This principle may lead to multidimensional memories, where the capacity per neuron is considerably increased with accurate detection of spike synchronization.
What problem does this paper attempt to address?