US 12,488,263 B2
Methods and apparatus for time-series forecasting using deep learning models of a deep belief network with quantum computing
Sherif Barrad, Ile-Bizard (CA); Ricardo A. Collado, Melrose, MA (US); Biren Agnihotri, Ontario (CA); and Olumide Akinola, North York (CA)
Assigned to Ernst & Young LLP, Toronto (CA)
Filed by Ernst & Young LLP, Toronto (CA)
Filed on Apr. 14, 2023, as Appl. No. 18/300,707.
Prior Publication US 2024/0403671 A1, Dec. 5, 2024
Int. Cl. G06N 7/01 (2023.01); G06N 3/08 (2023.01); G06N 3/084 (2023.01); G06N 3/088 (2023.01); G06N 5/01 (2023.01); G06N 20/00 (2019.01); G06N 20/20 (2019.01)
CPC G06N 7/01 (2023.01) [G06N 3/08 (2013.01); G06N 3/084 (2013.01); G06N 3/088 (2013.01); G06N 5/01 (2023.01); G06N 20/20 (2019.01); G06N 20/00 (2019.01)] 20 Claims
OG exemplary drawing
 
1. An apparatus, comprising:
a processor; and
a memory operatively coupled to the processor, the memory storing instructions to cause the processor to:
receive input data for a Deep Belief Network (DBN) that includes an indication of a time-series problem;
randomly initialize a plurality of weights and a first optimization function for a first deep learning model from a plurality of deep learning models associated with the DBN, the plurality of weights representing a strength between visible units representing the input data and hidden units of the first deep learning model;
generate, via the first deep learning model and using the input data and a subset of weights from the plurality of weights, a representation of the input data, the subset of weights, the input data, and the representation of the input data to be transmitted to a quantum compute device;
iteratively perform, until an error value associated with the DBN is below a predetermined threshold:
receive a plurality of sampled values from the quantum compute device using a second optimization function associated with the quantum compute device, the plurality of sampled values generated using the subset of weights, the input data, and the representation of the input data;
update, based on the plurality of sampled values, the subset of weights to train the first deep learning model to produce a trained deep learning model, the trained deep learning model configured to generate an updated representation of the input data;
generate, via a regression layer associated with the DBN, output data based on the updated representation of the input data;
iteratively update, via backpropagation of the regression layer, a set of weights of the regression layer to reduce the error value associated with the DBN; and
reconstruct, via the first deep learning model, the representation of the input data based on the set of weights updated by the regression layer, to produce a reconstructed representation of the input data.