Fast unsupervised learning for visual pattern recognition using spike timing dependent plasticity

Abstract

Real-time learning needs algorithms operating in a fast speed comparable to human or animal, however this is a huge challenge in processing visual inputs. Research shows a biological brain can process complicated real-life recognition scenarios at milliseconds scale. Inspired by biological system, in this paper, we proposed a novel real-time learning method by combing the spike timing-based feed-forward spiking neural network (SNN) and the fast unsupervised spike timing dependent plasticity learning method with dynamic post-synaptic thresholds. Fast cross-validated experiments using MNIST database showed the high e�ciency of the proposed method at an acceptable accuracy

Unsupervised AER Object Recognition Based on Multiscale Spatio-Temporal Features and Spiking Neurons

“…It can be seen that the distribution of features is heavily skewed. Linear coding functions are used by many existing works [10], [28] to convert these features to spikes for simplicity, but such functions cannot change the distribution of data and thus the temporal distribution of feature spikes are still skewed. This skewed distribution of feature spikes will lead to two problems: 1) higher-response features have less impact on recognition process.…”

“…The detailed settings of Gabor filters are listed in TABLE II. These parameter settings have been proved solid on the task of visual feature capturing, and inherited in many works [10], [12], [28]. The time constant of feature response τ leak is set according to the time length of the symbol in each dataset.…”

“…We compare the performance of the approach using conventional linear coding function [10], [28] and the proposed natural logarithm coding function. The linear coding function is set as follows: t spike = −ar+b, where a = t w /r max and b = t w .…”

Lattice map spiking neural networks (LM-SNNs) for clustering and classifying image data

“…The work by Diehl & Cook [14] uses a simple three-layer spiking neural network to classify the MNIST handwritten digits by learning synaptic weights in an unsupervised fashion with several different STDP rules. A number of other previous projects use STDP to train SNNs to classify image data [15,16]. The former uses Gabor filters as a pre-processing input step to detect simple features, which are then used as input to their network [15].…”

“…A number of other previous projects use STDP to train SNNs to classify image data [15,16]. The former uses Gabor filters as a pre-processing input step to detect simple features, which are then used as input to their network [15]. They use a rank-order encoding scheme on the input spikes, which are processed through the network, in which the winner-take-all classification strategy is used on its output activity.…”