Quantum Experiments Open is an open-source initiative that aims to bring the exciting frontier between quantum computing and artificial intelligence closer to the research and developer community. At its core, the project demonstrates how variational quantum classifiers (VQCs) can be implemented and trained to distinguish between two classes of data—in the current example, benign versus malicious samples—while providing a reproducible framework that highlights the potential of hybrid quantum–classical algorithms in AI and cybersecurity.
What makes this project particularly relevant is its seamless integration with Amazon Braket, AWS’s fully managed quantum computing service. By relying on Braket, users can move smoothly from testing locally to running experiments at scale in the cloud. The reference implementation provided in the repository uses the SV1 state vector simulator, a powerful simulator available on Braket that supports up to 34 qubits. This choice allows the community to work with realistic quantum workloads without needing direct access to physical quantum hardware, while still using the same APIs that make the transition to real devices straightforward.
The algorithm follows a clear workflow: the quantum circuit parameters are initialized with small random values, optimized with the Adam optimizer to minimize binary cross-entropy loss, evaluated with metrics such as accuracy, confusion matrix, and classification report, and finally visualized to show how the training loss evolves. These steps can be executed on Braket notebooks (for instance, using a ml.t3.medium instance) and then dispatched to SV1 for simulation, combining the convenience of managed infrastructure with the scalability of quantum resources in the cloud.
Another strength of Quantum Experiments Open is its flexibility. The repository provides two variants of the same VQC algorithm:
A complete version, which runs longer and processes larger batches to provide more faithful results.
A fast version, optimized for quick iterations, using fewer epochs and smaller datasets—ideal for testing directly in Braket notebooks before committing to full-scale runs.
This dual approach makes it possible to balance speed and fidelity: developers can iterate rapidly on Braket with the fast version, and once satisfied, scale up to the complete version for deeper evaluation.
The project is fully open source under the MIT license, and contributions from the community are welcome. While the repository originates in Catalan, contributions in English or any language are encouraged, and users are free to adapt the code for their own experiments and applications.
You can find the repository here: https://github.com/gcjordi/quantumexperimentsopen/wiki
By showcasing how variational quantum classifiers can be trained and evaluated directly on Amazon Braket, Quantum Experiments Open provides both a practical introduction to hybrid quantum–classical workflows and a solid foundation for further exploration of quantum machine learning in the context of artificial intelligence and cybersecurity.
Top comments (0)