US 11,989,162 B2
System and method for optimized processing of information on quantum systems
Jai Ganesh, Bangalore (IN); Udayaadithya Avadhanam, Bangalore (IN); Nachiket Kare, Nashik (IN); Ashutosh Vyas, Jaipur (IN); Rajendrakumar Premnarayan Mishra, Thane (IN); and Rohit Kumar Patel, Bangalore (IN)
Assigned to Mphasis Limited, (IN)
Filed by Mphasis Limited, Bangalore (IN)
Filed on Sep. 17, 2021, as Appl. No. 17/478,287.
Claims priority of application No. 202041040408 (IN), filed on Sep. 18, 2020.
Prior Publication US 2022/0092035 A1, Mar. 24, 2022
Int. Cl. G06N 20/00 (2019.01); G06F 16/178 (2019.01); G06N 10/00 (2022.01)
CPC G06F 16/1794 (2019.01) [G06N 10/00 (2019.01); G06N 20/00 (2019.01)] 38 Claims
OG exemplary drawing
 
1. A system for improved representation of classical data on quantum systems, wherein the system comprises:
a memory storing program instructions;
a processor executing program instructions stored in the memory; and
a feature definition engine configured to receive input classical data and create a feature set from the classical data; a feature space transformation engine configured to:
perform a functional transformation on the created feature set to reduce high dimensional data associated with the feature set and generate a low dimensional feature space dataset without loss of information; and
perform a feature space transformation on the low-dimensional feature space dataset to obtain a new feature space dataset with enhanced feature representation of the low-dimensional feature space dataset in a multi-dimensional space, wherein the new feature space dataset results in optimal mapping of the input classical data into a quantum format;
a batch preparation and selection engine configured to optimally sample the new feature space dataset and select batches of the sampled dataset; and
a quantum prediction engine configured to map the received batches of sampled dataset into an optimized quantum format for loading the sampled dataset into quantum states, wherein the quantum prediction engine comprises:
a feature engineering engine configured to convert an output quantum measurement data generated by a computation associated with the mapping by quantum circuits into a data format for feeding into a deep neural network of the quantum prediction engine, the converted data is passed through a fully connected layer of the deep neural network to obtain a classical output data; and
a feature space evaluation engine configured to evaluate performance of the system and provide feedback to the feature definition engine and the feature space transformation engine continuously for iteratively refining and redefining the obtained new feature space dataset for efficient quantum predictive tasks.