ABSTRACT
With inherent sparse spike-based coding and asynchronous event-driven computation, spiking neural network (SNN) is naturally suitable for processing event stream data of event cameras. In order to improve the feature extraction and classification performance of bio-inspired hierarchical SNNs, in this paper an event camera object recognition system based on biological synaptic plasticity is proposed. In our system input event streams were firstly segmented adaptively using spiking neuron potential to improve computational efficiency of the system. Multi-layer feature learning and classification are implemented by our bio-inspired hierarchical SNN with synaptic plasticity. After Gabor filter-based event-driven convolution layer which extracted primary visual features of event streams, we used a feature learning layer with unsupervised spiking timing dependent plasticity (STDP) rule to help the network extract frequent salient features, and a feature learning layer with reward-modulated STDP rule to help the network learn diagnostic features. The classification accuracies of the network proposed in this paper on the four benchmark event stream datasets were better than the existing bio-inspired hierarchical SNNs. Moreover, our method showed good classification ability for short event stream input data, and was robust to input event stream noise. The results show that our method can improve the feature extraction and classification performance of this kind of SNNs for event camera object recognition.
Subject(s)
Visual Perception , Learning , Action Potentials , Neural Networks, Computer , Neuronal PlasticityABSTRACT
Long-term potentiation and long-term depression are enduring changes in synaptic strength, induced by specific patterns of synaptic activity, that have received much attention as cellular models of information storage in the central nervous system. Work in a number of brain regions, from the spinal cord to the cerebral cortex, and in many animal species, ranging from invertebrates to humans, has demonstrated a reliable capacity for chemical synapses to undergo lasting changes in efficacy in response to a variety of induction protocols. In addition to their physiological relevance, long-term potentiation and depression may have important clinical applications. A growing insight into the molecular mechanisms underlying these processes, and technological advances in non-invasive manipulation of brain activity, now puts us at the threshold of harnessing long-term potentiation and depression and other forms of synaptic, cellular and circuit plasticity to manipulate synaptic strength in the human nervous system. Drugs may be used to erase or treat pathological synaptic states and non-invasive stimulation devices may be used to artificially induce synaptic plasticity to ameliorate conditions arising from disrupted synaptic drive. These approaches hold promise for the treatment of a variety of neurological conditions, including neuropathic pain, epilepsy, depression, amblyopia, tinnitus and stroke.