Storage capacity of phase-coded patterns in sparse neural networks

¹ Dipartimento di Fisica “E. R. Caianiello”, Università di Salerno - Salerno, Italy, EU
² INFN, Sezione di Napoli e Gruppo Collegato di Salerno - Italy, EU
³ Dipartimento di Scienze Fisiche, Università di Napoli Federico II - Napoli, Italy, EU
⁴ CNR-SPIN, Unità di Napoli - Napoli, Italy, EU

^a antonio.decandia@na.infn.it

Received: 18 March 2011
Accepted: 1 June 2011

Abstract

We study the storage of multiple phase-coded patterns as stable dynamical attractors in recurrent neural networks with sparse connectivity. To determine the synaptic strength of existent connections and store the phase-coded patterns, we introduce a learning rule inspired to the spike-timing–dependent plasticity (STDP). We find that, after learning, the spontaneous dynamics of the network replays one of the stored dynamical patterns, depending on the network initialization. We study the network capacity as a function of topology, and find that a small-world–like topology may be optimal, as a compromise between the high wiring cost of long-range connections and the capacity increase.