Beijing – September 19, 2026 -- WiMi Hologram Cloud Inc. (NASDAQ: WiMi) has developed a reinforcement learning-based framework to automate the design of quantum encoding circuits, aiming to cut the resource costs and inefficiencies that have limited quantum machine learning (QML) model performance.
WiMi replaces manual circuit design with automated search architecture
The new scheme uses reinforcement learning to generate task-specific encoding circuit architectures automatically, moving away from the heuristic, manually designed approaches that have dominated QML development. WiMi states the framework targets three persistent industry problems: poor task adaptability, low search efficiency, and insufficient handling of multiple design objectives simultaneously.
Model-based approach cuts quantum hardware evaluation requirements
Unlike conventional reinforcement learning methods that require testing every candidate circuit on actual quantum hardware, WiMi's system builds an environment model to predict circuit performance directly. This eliminates the need for exhaustive hardware evaluations, reducing quantum resource consumption and shortening the search process for optimal circuit architectures under constrained computing resources.
Hierarchical structure narrows search space to boost stability
WiMi's design breaks encoding circuits into multiple levels of basic modules rather than searching complete circuit architectures at once. A reinforcement learning agent explores and combines these modules level by level, progressively constructing a full circuit adapted to specific tasks. The company says this reduces search dimensions, avoids redundant computation, and improves the structural rationality and performance stability of the final circuit compared with traditional search algorithms.
Multi-objective reward function balances performance against noise robustness
The framework incorporates a multi-objective reward function that allows simultaneous optimization of model performance, quantum resource consumption, and noise robustness during circuit generation. WiMi says this synergistic optimization avoids the performance imbalances that arise when systems are tuned for a single objective, allowing the framework to be matched to different application scenarios.
WiMi commits to further development of QML technology
WiMi said it will continue researching the intersection of quantum computing and machine learning, refining the encoding circuit generation solution as part of broader work in the QML field.