Lena Tokuhiro | Searcharxiv

arXiv2026

The QuaST Decision Tree: Achieving Automation With Data-Based Recommendations

Quantum computers are increasingly powerful. Software tools for the development of quantum-enhanced algorithms are maturing. However, the software stack still lacks the connection to applications that would enable hybrid algorithms combining classical and quantum computing steps. End users need to be assisted in choosing the best combination of preprocessing, postprocessing, classical and quantum algorithms options. The application-facing software stack is therefore required to cover problem modeling, encoding, algorithm selection and hyperparameter tuning. A variety of tools exist for specific recommendations. The QuaST Decision Tree reflects the complexity in combining individual decisions in its modular network structure, consisting of flexible computation nodes with modular recommendations. It can easily be configured to serve in an industrial solver, an HPC software stack, or for rapid prototyping in development. The key ingredient, automation, is delivered by modules. We present one such module judging the feasibility of variational algorithms based on a robust scalability analysis and classification of problem instances. The automation improves the performance of end-to-end solutions, highlights the benefit to be gained from the hybrid quantum solution, reduces expensive trial-and-error testing, and leads to an improved utilization of quantum devices for a practical benefit.

quant-ph↗

arXiv2026

Discovering Data Encoding Strategies for Quantum-Classical Neural Networks Using Monte Carlo Tree Search

Quantum machine learning (QML) has attracted considerable research interest, yet whether it offers practical benefits over classical approaches remains an open question. The choice of data encoding significantly influences QML performance, but why certain encodings outperform others remains poorly understood. We employ Monte Carlo Tree Search (MCTS) to discover optimal data encoding circuits for a quantum-classical convolutional neural network (QCCNN) combining a non-variational quantum block for feature extraction with a classical classifier. Evaluating on two medical imaging datasets, the discovered circuits outperform commonly used encoding strategies while showing competitive results compared to purely classical counterparts. We further analyze metrics to identify predictors of encoding performance. Entanglement capability and Fourier decomposition provide minimal insight, whereas the effective rank of the feature maps exhibits meaningful correlation and can serve as a threshold criterion to accelerate the search for high-performing encodings. Our findings provide both a practical method for encoding discovery and new insights into what makes data encodings effective in QML.

quant-ph↗