International Conference on Algorithms, Methodology, Models and Applications in Emerging Technologies (ICAMMAET)

28 / 2017-01-10 20:25:03

A handwritten numeral recognition method based on STDP based with unsupervised learning

Spiking neural network;spike-timing-dependent-plasticity;numeral recognition;unsupervised learning;

In order to sovle the problems of SNN lacking of biologically plausible mechnisms and performance,we present a SNN for numeral recognition based on mechanisms with increased biological plausibility,i.e.,Using unsupervised learning based on conductance rather than current-based synapses,lateral inhibition and adaptive spike thresholds.Experimental results show that the method in this paper has significant advantages in recognition accuracy in MNIST.It not only increases the accuracy of recognition,but also increases recognition efficiency over which uses BP neural network.