Nonuniform magnetic domain-wall synapses enabled by population coding

Traditional artificial intelligence implemented in software is usually executed on accurate digital computers. Nevertheless, the nanoscale devices for implementation of neuromorphic computing may not be ideally identical, and performance reduced by nonuniform devices. In biological brains, information encoded a cluster neurons such that variability nerve cells does influence accuracy human cognition movement. Here, we introduce population encoding strategy demonstrate this can overcome problems caused Using magnetic memristor device based current-induced domain-wall motion as an example, show imperfect storage applied hardware network to perform principal component analysis (PCA), unsupervised classification comparable conventional PCA using synaptic weights. Our results pave way lower criteria uniformity