R&D: Neuromorphic Computing Using Non-Volatile MemoryDense crossbar arrays of non-volatile memory
By Francis Pelletier on 2016.12.12
Advances in Physics: X has published an article written by Geoffrey W. Burr, Robert M. Shelby, IBM Research Almaden, San Jose, CA, Abu Sebastian, IBM Research Zurich, Ruschlikon, Switzerland, Sangbum Kim, Seyoung Kim, IBM T.J. Watson Research Center, Yorktown Heights, NY, Severin Sidler, EPFL, Lausanne, Switzerland, Kumar Virwani, IBM Research Almaden, San Jose, CA, Masatoshi Ishii, IBM Tokyo Research Laboratory, Tokyo, JP, Pritish Narayanan, Alessandro Fumarola, Lucas L. Sanches, IBM Research Almaden, San Jose, CA, Irem Boybat, Manuel Le Gallo, IBM Research Zurich, Ruschlikon, Switzerland, Kibong Moon, Jiyoo Woo, Hyunsang Hwang, Department of Material Science an Engineering, Pohang University of Science and Technology, Pohang, Korea, and Yusuf Leblebici, EPFL, Lausanne, Switzerland.
Abstract: "Dense crossbar arrays of non-volatile memory (NVM) devices represent one possible path for implementing massively-parallel and highly energy-efficient neuromorphic computing systems. We first review recent advances in the application of NVM devices to three computing paradigms: spiking neural networks (SNNs), deep neural networks (DNNs), and 'Memcomputing'. In SNNs, NVM synaptic connections are updated by a local learning rule such as spike-timing-dependent-plasticity, a computational approach directly inspired by biology. For DNNs, NVM arrays can represent matrices of synaptic weights, implementing the matrix–vector multiplication needed for algorithms such as backpropagation in an analog yet massively-parallel fashion. This approach could provide significant improvements in power and speed compared to GPU-based DNN training, for applications of commercial significance. We then survey recent research in which different types of NVM devices - including phase change memory, conductive-bridging RAM, filamentary and non-filamentary RRAM, and other NVMs - have been proposed, either as a synapse or as a neuron, for use within a neuromorphic computing application. The relevant virtues and limitations of these devices are assessed, in terms of properties such as conductance dynamic range, (non)linearity and (a)symmetry of conductance response, retention, endurance, required switching power, and device variability."