Learning Real-World Stimuli in a Neural Network with Spike-Driven Synaptic Dynamics

TLDR

The authors propose a spike‑driven synaptic plasticity model to enable an electronic hardware device that can semisupervisedly classify complex stimuli. The model comprises integrate‑and‑fire neurons with constant leak and a floor, bistable synapses that change upon presynaptic spikes with a sign determined by postsynaptic depolarization and a variable integrating postsynaptic action potentials, and training applies sequential input patterns with an instructor signal that is removed afterward, leaving output driven solely by input activity. The network retains synaptic states indefinitely, resists spontaneous activity, correctly classifies thousands of overlapping patterns with performance comparable to or exceeding artificial neural networks, and its dynamics align with experimental observations of spike‑timing‑dependent plasticity.

Abstract

We present a model of spike-driven synaptic plasticity inspired by experimental observations and motivated by the desire to build an electronic hardware device that can learn to classify complex stimuli in a semisupervised fashion. During training, patterns of activity are sequentially imposed on the input neurons, and an additional instructor signal drives the output neurons toward the desired activity. The network is made of integrate-and-fire neurons with constant leak and a floor. The synapses are bistable, and they are modified by the arrival of presynaptic spikes. The sign of the change is determined by both the depolarization and the state of a variable that integrates the postsynaptic action potentials. Following the training phase, the instructor signal is removed, and the output neurons are driven purely by the activity of the input neurons weighted by the plastic synapses. In the absence of stimulation, the synapses preserve their internal state indefinitely. Memories are also very robust to the disruptive action of spontaneous activity. A network of 2000 input neurons is shown to be able to classify correctly a large number (thousands) of highly overlapping patterns (300 classes of preprocessed Latex characters, 30 patterns per class, and a subset of the NIST characters data set) and to generalize with performances that are better than or comparable to those of artificial neural networks. Finally we show that the synaptic dynamics is compatible with many of the experimental observations on the induction of long-term modifications (spike-timing-dependent plasticity and its dependence on both the postsynaptic depolarization and the frequency of pre- and postsynaptic neurons).