Methods and apparatus for time-series forecasting using deep learning models of a deep belief network with quantum computing

A non-transitory, processor-readable medium storing instructions that executed by a processor, cause the processor to:

• receive input data for a Deep Belief Network (DBN) indicating a time-series problem;
• randomly initialize, based on the input data, a plurality of weights for each deep learning model from a plurality of deep learning models associated with the DBN, the plurality of weights representing a strength between the input data and hidden units of the deep learning model;
• iteratively perform, until an error value associated with the DBN is below a predetermined threshold:
• receiving, from a quantum compute device using an optimization function associated with the quantum compute device, a plurality of sampled values, the plurality of sampled values generated using at least a subset of weights from the plurality of weights and associated with a deep learning model from the plurality of deep learning models;
• updating, based on the plurality of sampled values, the subset of weights;
• training the deep learning model based on the updated subset of weights, to produce a trained deep learning model, the trained deep learning model configured to generate an updated representation of the input data;
• generating, via a regression layer associated with the DBN, output data based on the updated representation of the input data;
• iteratively updating, via backpropagation of the regression layer, a set of weights of the regression layer to reduce the error value;
• and
• reconstructing, via the deep learning model, the representation of the input data based on the subset of weights updated by the regression layer, to produce a reconstructed representation of the input data.

The non-transitory, processor-readable medium of 17, wherein:

• the deep learning model is a first deep learning model, the trained deep learning model is a first trained deep learning model, the subset of weights is a first subset of weights, the representation of the input data is a first representation of the input data, the plurality of sampled values is a first plurality of sampled values, the reconstruction of the representation of the input data is a first reconstruction of the representation of the input data, and
• the instructions to cause the processor to iteratively perform, until the error value is below the predetermined threshold, further includes:
• receiving, from the compute device using the optimization function, a second plurality of sampled values, the second plurality of sampled values generated based on at least a second subset of weights from the plurality of weights and the first reconstructed representation of the input data, the second subset of weights associated with a second deep learning model from the plurality of deep learning models;
• updating, based on the second plurality of sampled values, the second subset of weights;
• training the second deep learning model based on the updated second subset of weights, to produce a second trained deep learning model, the second trained deep learning model configured to generate a second updated representation of the input data;
• generating, via the regression layer, the output data based on the second updated representation of the input data;
• and
• iteratively updating, via backpropagation of the regression layer, a set of weights of the regression layer to reduce the error value.

A method, comprising:

• receiving input data for a Deep Belief Network (DBN) that includes an indication of a time-series problem;
• randomly initializing, based on the input data, a plurality of weights for a deep learning model from a plurality of deep learning models associated with the DBN, the plurality of weights representing a strength between the input data and hidden units of the deep learning model;
• generating, via the 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;
• receiving a plurality of sampled values from the quantum compute device using an 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;
• updating, based on the plurality of sampled values, the subset of weights to train the 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;
• and
• generating, via a regression layer associated with the DBN, output data based on the updated representation of the input data.