Systems and methods for fast and repeatable embedding of high-dimensional data objects using deep learning with power efficient GPU and FPGA-based processing platforms

A system for generating a low-dimensional summary vector representation of a high-dimensional data object in a computer memory, comprising:

• a first processor in communication with a first computer readable memory;
• an object embedding module embodied in the first computer readable memory, wherein the object embedding module, when executed by the first processor, creates an embedding of a plurality of high-dimensional training data objects, each of the plurality of high-dimensional training data objects comprising a different representation of an actual object, where the embedding comprises a set of ordered pairs, each ordered pair comprising one of the plurality of high-dimensional training data objects and a corresponding low-dimensional training vector created by a selected embedding algorithm operating within the object embedding module;
• a second processor in communication with a second computer readable memory;
• a deep architecture training module embodied in the second computer readable memory, wherein the deep architecture training module, when executed by the second processor, trains a neural network with the set of ordered pairs to produce a deterministic deep architecture function that can substantially replicate the embedding;
• a third processor in communication with a third computer readable memory;
• and
• a deep architecture deployment module embodied in the third computer readable memory, wherein the deep architecture deployment module, when executed by the third processor:
• receives a high dimensional input data object from an external data source, said high-dimensional input data object obtained from an observation of a physical object, and
• invokes the deep architecture function to generate a low-dimensional summary vector representation of the received high-dimensional input data object.