公开(公告)号：GB2581731B

公开(公告)日：2022-11-09

申请号：GB202006982

申请日：2018-10-23

Applicant: IBM

Inventor： MANUEL LE GALLO-BOURDEAU ,  ABU SEBASTIAN ,  IREM BOYBAT KARA ,  EVANGELOS STAVROS ELEFTHERIOU ,  NANDAKUMAR SASIDHARAN RAJALEKSHMI

IPC: G06N3/08 ,  G06N3/063

Abstract: Methods and apparatus are provided for training an artificial neural network having a succession of layers of neurons interposed with layers of synapses. A set of crossbar arrays of memristive devices, connected between row and column lines, implements the layers of synapses. Each memristive device stores a weight for a synapse interconnecting a respective pair of neurons in successive neuron layers. The training method includes performing forward propagation, backpropagation and weight-update operations of an iterative training scheme by applying input signals, associated with respective neurons, to row or column lines of the set of arrays to obtain output signals on the other of the row or column lines, and storing digital signal values corresponding to the input and output signals. The weight-update operation is performed by calculating digital weight-correction values for respective memristive devices, and applying programming signals to those devices to update the stored weights.

公开(公告)号：GB2581731A

公开(公告)日：2020-08-26

申请号：GB202006982

申请日：2018-10-23

Applicant: IBM

Inventor： MANUEL LE GALLO-BOURDEAU ,  ABU SEBASTIAN ,  IREM BOYBAT KARA ,  EVANGELOS STAVROS ELEFTHERIOU ,  NANDAKUMAR SASIDHARAN RAJALEKSHMI

IPC: G06N3/08 ,  G06N3/063

Abstract: Methods and apparatus are provided for training an artificial neural network having a succession of layers of neurons interposed with layers of synapses. A set of crossbar arrays of memristive devices, connected between row and column lines, implements the layers of synapses. Each memristive device stores a weight Ŵ for a synapse interconnecting a respective pair of neurons in successive neuron layers. The training method comprises performing forward propagation, backpropagation and weight-update operations of an iterative training scheme by, in at least one of the forward propagation and backpropagation operations of the scheme, applying input signals, associated with respective neurons, to one of row and column lines of the set of arrays to obtain output signals on the other of the row and column lines, and storing digital signal values corresponding to the input and output signals in a digital processing unit operatively coupled to the set of arrays. The weight-update operation of the scheme is performed by calculating, in the digital processing unit, digital weight-correction values ΔW, dependent on the stored digital signal values, for respective memristive devices, and applying programming signals to those devices to update the stored weights Ŵ in dependence on the respective digital weight-correction values ΔW.