European Institutions developing neural networks for quantum error correction

European quantum computing research groups launch project to establish and commercialize a modern approach to quantum control based on Neural Networks
15 April 2022

With 2022 expected to see more use cases of quantum computing, more organizations are looking to implement the technology in their organization. While quantum computing sounds exciting, the reality is, that it is not as easy as it schemes.

In fact, researchers believe that there is not only still a lot to be discovered in quantum computing but also several challenges that need to be addressed. This is why in most cases; quantum computing is still being used to trial use cases instead of being implemented. To work with quantum computing would require a comprehensive hardware and software platform that is capable of performing the most complex quantum algorithms and experiments.

And this is where the Quantum Orchestration Platform (QOP) comes in. Created by Quantum Machines, the platform fundamentally redefines the architecture of the quantum control stack for the control and operations of quantum processors. The full-stack hardware and software platform is capable of running even the most complex algorithm.

This includes quantum error correction, multi-qubit calibration, and more. Helping achieve the full potential of any quantum processor, the QOP allows for unprecedented advancement and speed-up of quantum technologies as well as the ability to scale into the thousands of qubits.

Despite this, there are still two main challenges in quantum computing –quantum error correction and optimal control. In order to overcome these challenges, Quantum Machines and Alice&Bob, a leading European developer of quantum processors together with top European quantum computing research groups have launched a project to establish and commercialize a radically new approach to quantum control based on Neural Networks.

The three-year project will focus on the development of a quantum controller that incorporates real-time neural networks capable of generating controls. The use of neural networks is expected to enhance the accuracy and performance of quantum processors, and heavily reduce the classical control resources needed, which is a true bottleneck towards scaling up error correction and optimal control methods.

The expected outcomes of the project are:

  • The deployment of a universal quantum controller with a user-friendly interface and accompanying open-source code libraries for the implementation of the new approach on a variety of quantum processors and devices.
  • The public availability of a cloud-based quantum processor with a unique user interface allows for the programming and execution of a rich variety of real-time neural networks. This will allow researchers to explore this new approach toward practical quantum computing and quantum sensing, even if they do not have direct access to quantum hardware.

According to Dr. Yonatan Cohen, CTO of Quantum Machines, the future viability of practical quantum computing is heavily dependent on achieving error correction consistently and efficiently.

“We expect the neural networks being developed as part of ARTEMIS to help improve our control over larger numbers of qubits, even in the face of environmental decoherence, to help facilitate the real-world deployment of quantum computers,” commented Dr. Cohen.

For Dr. Théau Peronnin, CEO of Alice&Bob, the company’s roadmap is predicated on a lean inspiration. They aim to reduce the minimum quantum resources required to build a fault-tolerant quantum computer.

“By making control more efficient, ARTEMIS advances that philosophy outside the cryostat and brings the reality of practical quantum computing one step closer,” added Dr. Peronnin.

The project will utilize the combined expertise of the participating companies and institutions in the fields of microwave engineering, machine learning, control theory, experimental quantum physics, commercial product design and realization, and industrial level quantum computers to realize the full potential of this project.

“We expect neural networks to help identify new strategies for quantum control”, said Benjamin Huard, Professor at Ecole Normale Supérieure de Lyon in France. “In particular, we expect a sizable improvement for discovering the optimal control laws in imperfect experimental settings. We are excited to gather such a strong consortium to test these ideas experimentally and build useful tools for quantum computing.”

The ARTEMIS project aims at establishing and commercializing a radically new neural-networks-based quantum control approach. It will use reinforcement learning on real-time experimental observations in order to overcome today’s main challenges in quantum computing – quantum error correction and optimal control.