Search International and National Patent Collections

1. (WO2018222896) GRADIENT-BASED TRAINING ENGINE FOR QUATERNION-BASED MACHINE-LEARNING SYSTEMS

Pub. No.:    WO/2018/222896    International Application No.:    PCT/US2018/035431
Publication Date: Fri Dec 07 00:59:59 CET 2018 International Filing Date: Fri Jun 01 01:59:59 CEST 2018
IPC: G06N 99/00
Applicants: INTEL CORPORATION
MARTINEZ-CANALES, Monica Lucia
SINGH, Sudhir K.
SHARMA, Vinod
BHANDARU, Malini Krishnan
Inventors: MARTINEZ-CANALES, Monica Lucia
SINGH, Sudhir K.
SHARMA, Vinod
BHANDARU, Malini Krishnan
Title: GRADIENT-BASED TRAINING ENGINE FOR QUATERNION-BASED MACHINE-LEARNING SYSTEMS
Abstract:
A deep neural network (DNN) includes hidden layers arranged along a forward propagation path between an input layer and an output layer. The input layer accepts training data comprising quaternion values, outputs a quaternion-valued signal along the forward path to at least one of the hidden layers. At least some of the hidden layers include quaternion layers to execute consistent quaternion (QT) forward operations based on one or more variable parameters. A loss function engine produces a loss function representing an error between the DNN result and an expected result. QT backpropagation-based training operations include computing layer-wise QT partial derivatives, consistent with an orthogonal basis of quaternion space, of the loss function with respect to a QT conjugate of the one or more variable parameters and of respective inputs to the quaternion layers.