Clustering of user activities based on adaptive threshold spiking neural networks

Abstract

Spiking neural networks are utilized in solving hard computation problems in intelligent systems. Spiking neural networks have a high computational power due to the implicit employment of various parameters such as input times and values in addition to neuron threshold, synaptic delays, and weights in their structures. On the other hand, smart environment techniques are emergent science in this decade. Intelligent systems represented by spiking neural network models and smart environments represented by sensors readings are utilized in this research for clustering users' activities during some period of time. A new learning algorithm for spiking neural network based on adaptation of the internal neuron threshold is proposed. Threshold adaptation is employed to help a spiking neuron to fire the lowest number of output spikes and to preserve all information of the input spike train on the same time. Simulations show that the clustering algorithm has encouraging results.