Quantum Machine Learning: Software AutoQML Facilitates Industrial Deployment

Image recognition during laser cutting as a concrete application example for the use of quantum ML algorithms. © Trumpf Group

The autoQML team at a project meeting 2023. © Fraunhofer IAO

In the joint project AutoQML, the Fraunhofer Institutes IAO and IPA, together with seven industry partners, developed a software application of the same name. The open-source software AutoQML combines quantum computing and machine learning. This makes algorithms of quantum machine learning accessible without in-depth expertise.

How can companies harness the potentials of digitalization and remain competitive? The use of technologies like artificial intelligence can help maximize the benefits of digital transformation. Especially, machine learning (ML) already plays a significant role in the digitalization strategies of many companies, enabling more efficient processes and new business models. However, there is often a lack of skilled personnel. As a result, implementing ML solutions has so far been associated with high effort. From data acquisition to selecting the appropriate algorithms and optimizing training, detailed ML expertise is required.

The approach of automated machine learning (AutoML) addresses these challenges and simplifies the deployment of artificial intelligence. It automates, in particular, the selection of specific ML algorithms. Users need less familiarity with ML and can focus more on their core processes.

In this context, the innovation of quantum computing promises to establish new solutions that significantly improve the AutoML approach. Additionally, quantum computing offers the computational power often needed for AutoML.

New Approach: Quantum Computing Takes Machine Learning to a New Level

The joint project "AutoQML" built on this innovation and achieved two main goals: First, the new approach AutoQML was developed. It extends the AutoML principle with newly developed quantum ML algorithms. Second, quantum computing elevates the AutoML approach to a new level, as certain problems can be solved more efficiently and sustainably with quantum computing than with conventional algorithms.

Led by the Fraunhofer Institute for Industrial Engineering IAO, the developed open-source software AutoQML now provides developers with easier access to conventional and quantum ML algorithms. The developed quantum ML components and methods have been compiled into a toolkit and made available to development teams. This enables users to deploy machine learning and quantum machine learning and to develop automated hybrid total solutions.

In addition to the Fraunhofer Institute for Production Technology and Automation IPA, companies GFT Integrated Systems, USU GmbH, IAV GmbH Engineering Company for Auto and Traffic, KEB Automation KG, Trumpf, and Zeppelin GmbH participated in the project. The solutions developed were tested using concrete application cases from the automotive and manufacturing sectors.

Benchmarking Study Demonstrates the Potential of AutoQML

In the final benchmarking study, the project consortium compared its open-source software AutoQML with the best known classical and quantum methods. A key result of the study: The automated solutions of AutoQML perform at least as well as the best manually identified classical and quantum methods. This opens up opportunities for developers to experiment with their own application cases.

The open-source software represents an important step towards broader application of quantum machine learning in industry, which can sustainably enhance the competitiveness and innovation capacity of companies.

The further dissemination of the technology by the industry partners promotes the transfer of research-oriented high technology into a broad industrial environment and aims to significantly strengthen the industrial location of Germany. The scientific findings from the project have been published in several publications. The project was funded by the Federal Ministry for Economic Affairs and Climate Action (BMWK) for a duration of three years.