Air Force Research Laboratory Partners with QC Ware to Apply Quantum Machine Learning in Identifying Flight Patterns of Unmanned Aircraft

Two-phased research collaboration explores use of QC Ware’s quantum machine learning algorithm for multiple mission applications

The Air Force Research Laboratory (AFRL) and QC Ware are exploring the use of one of QC Ware’s proprietary quantum machine learning algorithms to understand the purpose or mission objective of unmanned aircraft by observing its flight path. Called q-means, the quantum algorithm for clustering and classification can also be applied to a variety of AFRL mission applications.

QC Ware has been an early partner in advancing a key AFRL objective — supporting quantum algorithm development in optimization, machine learning, and quantum simulation for implementation in near-term quantum computers. The project is part of a larger AFRL effort to engage with expert researchers in industry, academia, and the Department of Defense to apply quantum information science to Air Force and Space Force concerns and ensure they remain the most advanced and capable force in the world.

“AFRL is pleased to partner with QC Ware on the development of quantum machine learning algorithms. Early and continued investment in quantum software firmly aligns with AFRL’s quantum strategy. As quantum computing hardware continues to rapidly advance and become more practical, we believe that these types of algorithms will readily find applications to real-world Air Force scenarios,” said Dr. Mike Hayduk, Deputy Director, AFRL Information Directorate.

Recommended AI News: Shutterstock Integrates with Google Drive for Enterprise, Unlocking Direct Access to High-Quality Images Across Google Workspace

“QC Ware values our long-term partnership with AFRL. We believe quantum software is a mission-critical advantage for organizations and nations, and can be useful in the near-term,” said Matt Johnson, CEO, QC Ware. “QC Ware’s research leadership in machine learning algorithms supports AFRL’s exploration into how quantum applications can ensure U.S. defense and space organizations maintain their superior edge.”

In the project’s first phase, AFRL and QC Ware researchers identified use cases for testing and evaluating the q-means algorithm to classify the flight path behavior of  unmanned aircraft. It focused on designing and delivering software for a prototype q-means algorithm for clustering and classification that could be executed on commercially available quantum computers and quantum simulators, and would have the potential to provide improved performance over classical algorithms for a broad variety of other Air Force missions as well. The team used standard software and libraries in building the q-means software to enable AFRL researchers to run and modify the software themselves.

q-means is significantly faster than its classical counterpart, k-means. Both algorithms use the same technique in clustering similar data points together, and are especially useful when data scientists are trying to better understand similarities of items in their data set.

Recommended AI News: DeepCube Acquired by Nano Dimension to Drive Industry 4.0 Evolution

Compared with k-means, the q-means approach to machine learning scales much better with more complex, high-dimensional data sets. The greater the number of dimensions the finer the granularity, enabling richer tracking of flight paths and other behaviors. The q-means method is 10x faster than k-means when clustering data sets with 64 dimensions, 30x faster on 256 dimensions, and 100x faster on 1,000 dimensions.

In the project’s second phase, the research collaboration will focus on evaluating the performance of the algorithm, using real data and implementing the algorithm on currently available quantum hardware. The experiment’s objective is to determine how much quantum resources are needed in terms of qubit count and error rate to achieve performance speed-ups over classical algorithms for clustering and classification.

“AFRL’s ongoing investment in algorithm development and in benchmarking the performance of quantum machine learning algorithms over classical algorithms, using real data and real quantum hardware, is encouraging to quantum machine learning research efforts,” said Iordanis Kerenidis, Head of Quantum Algorithms-International, QC Ware. “We view our research collaboration with AFRL as an important opportunity to implement QC Ware’s landmark q-means algorithm to solve computational bottlenecks in key mission applications.”

Recommended AI News: Noosh Unveils Sustainability Scores to Give Marketers the Ability to Proactively Select Suppliers who are Reducing the Environmental Impact of Printed Marketing Materials