In this paper, we consider a novel approach to information representation in spiking neural networks. In a certain sense it is a combination of two well-known coding schemes - rate coding and temporal coding. Namely, it is based on asynchronous activity of ensembles of polychronous neuronal groups - groups of neurons firing in determined order with strictly fixed relative temporal delays. We demonstrate how a rate-coded input signal may be converted into this representation form by network structures which are formed as a result of network self-organization process based on STDP-style synaptic plasticity.