作者: Lubica Benuskova , Nikola Kasabov , Simei Gomes Wysoski
DOI: 10.1007/11840817_7
关键词: Pattern recognition 、 Face (geometry) 、 Synaptic plasticity 、 Artificial intelligence 、 Image processing 、 Facial recognition system 、 Machine learning 、 Pattern recognition (psychology) 、 Artificial neural network 、 Computation 、 Event (computing) 、 Computer science
摘要: This paper presents an on-line training procedure for a hierarchical neural network of integrate-and-fire neurons. The is done through synaptic plasticity and changes in the structure. Event driven computation optimizes processing speed order to simulate networks with large number applied face recognition task. Preliminary experiments on public available image dataset show same performance as optimized off-line method. A comparison other classical methods demonstrates properties system.