The algorithm demonstrates how to formulate quantum SISR as a sparse coding optimization problem and solve it using the Dynex neuromorphic computing platform via the Dynex SDK. This AQC-based algorithm has been shown to improve SISR accuracy.
Refer to the original dynex use case: https://github.com/dynexcoin/DynexSDK/tree/main/Quantum-SISR
Quantum Algorithm Contributors:
“Quantum Annealing for Single Image Super-Resolution” by Han Yao Choong, Suryansh Kumar, and Luc Van Gool (ETH Zurich) https://arxiv.org/abs/2304.08924
Knowledge background: This is a quantum computer algorithm
A well-known classical approach to SISR relies on a well-developed patch-wise sparse modeling of the problem. However, the current state of the field is that deep neural networks (DNNs) have shown far superior results than classical approaches. Nevertheless, quantum computing is expected to become increasingly prominent in machine learning problems soon. Among the two paradigms of quantum computing, namely universal gate quantum computing and adiabatic quantum computing (AQC), the latter has been successfully applied to practical computer vision problems, where quantum parallelism has been exploited to efficiently solve combinatorial optimization.
Quantum entanglement content assessment
The advantages offered by quantum computing come from one or more specific key steps in algorithm design, which include problems that are typically expensive to solve using only CPUs, such as integer factorization, graph cuts, unstructured search, and other important combinatorial optimization problems. To this end, applying quantum computing to such problems can bring commendable speedups over CPU computing by exploiting quantum parallelism (i.e., the ability to perform operations on exponentially many superposition memory states simultaneously).
Quantum entanglement detection
In modern quantum computing, there are two paradigms to solve different problems suitable for quantum parallelism: universal gate quantum computing and adiabatic quantum computing (AQC). In this paper, we focus on the adiabatic quantum computing paradigm. In our formulation, sparse coding is used as a fundamental computational problem suitable for quantum computing.
When a QUBO problem is submitted through the D-Wave Leap interface, it takes some time to communicate, schedule, and assign the problem to a specific QPU. Subsequently, the QUBO problem is programmed onto the QPU, annealed, sampled, and then post-processed. In this case, the QPU optimization runtime refers to the total amount of time spent on programming, annealing, and sampling on the QPU, while the AQC preparation runtime refers to the total amount of time spent on tasks such as communication, scheduling, and dispatching. Allocation, post-processing, and other overheads are done by the server-side CPU.
This study migrated to Dynex Effects
