Design of Neuromorphic Architectures with Memristors

Advent of Spintronics in Neuromorphic Computing Systems: A Study

“…The spintronic storage elements are shared over a logical circuit plane in a novel logic large scale integration architecture and are propitious to curb power wall, as well as performance wall, put down to current logic large scale integration based on only CMOS, this architecture can be termed as spintronics based non-volatile logic-inmemory architecture [79]. The logic-in-memory, along with dynamically reconfigurable logic composed of LUT's and flipflops, all spin logic device with built-in memory, neuromorphic with memristor synapses crossbar configuration, and threshold logic gates are considered to be normally off computing systems due to negligible static leakage and they are different verticals of spintronics-based computing architectures and logic circuits [80]. At the Nanoscale level which is very small size, the CMOS compatible nanodevices can mimic the biological synapses, and the research effort is been carried out to invent such CMOS compatible spintronics nanodevices with computing with the onsite non-volatile magnetic memory could be one of the advantages with CMOS compatible spintronics nanodevices and combining at nanoscale being the other one, which could allow on-site storing of system parameters such as synaptic weights of the neural network [81].…”

Unsupervised learning in neuromemristive systems

“…The utility of these systems has been demonstrated in various application domains, especially image processing/analysis. See [2] for a review. Learning in these systems has primarily been supervised.…”