WiMi Hologram Cloud Inc. (NASDAQ: WiMi) is extending its technological reach beyond augmented reality by applying machine learning to a critical area of secure communication: quantum key distribution. The company reports training and evaluating three neural network models, Backpropagation (BPNN), Radial Basis Function (RBFNN), and Generalized Regression (GRNN), to optimize parameters within its dual-field quantum key distribution (TF-QKD) system. WiMi found all three models could accurately predict optimal parameters, with the Radial Basis Function Neural Network (RBFNN) highlighted as particularly suitable for high-dimensional data and scenarios requiring high precision. This research demonstrates a comparative approach to reducing computational demands in quantum cryptography, potentially accelerating key generation and improving system responsiveness.
WiMi Hologram Cloud Inc. is extending its research from augmented reality into quantum key distribution (QKD), specifically exploring how machine learning can optimize the complex parameters of dual-field quantum key distribution (TF-QKD) systems. This diversification for the NASDAQ-listed company focuses on advanced communications and cybersecurity technologies, leveraging artificial intelligence to address challenges within quantum cryptography. The investigation centers on utilizing neural networks to predict optimal parameter configurations, aiming to reduce the computational burden traditionally associated with TF-QKD.
According to the company, each model demonstrated an ability to accurately predict optimal parameters, though with varying degrees of efficiency. BPNN, due to its relatively simple structure, had the fastest computation speed, while RBFNN and GRNN, though slightly more complex in terms of computational cost, remained within acceptable limits, and their enhanced prediction accuracy often provided more practical application value. This comparative analysis highlights a trade-off between speed and precision, crucial considerations for real-world implementation. The company’s findings suggest that BPNN is best suited for applications demanding rapid response times with lower precision, while RBFNN or GRNN are preferable when high accuracy is paramount, even at the cost of increased processing demands. WiMi states that for scenarios requiring rapid response with lower precision demands, BPNN is the ideal choice, further clarifying the nuanced performance characteristics of each model.
The technical advantage, as WiMi explains, lies in the ability of neural networks to dynamically adapt to changing quantum communication environments and accelerate key generation rates. Future research will focus on incorporating more advanced architectures like deep learning and reinforcement learning, with the ultimate goal of creating more efficient and intelligent quantum key distribution systems and contributing to the development of secure quantum communication networks.
Compared to LSA, the neural network-based prediction method achieved a significant reduction in computation time, cutting it by multiple orders of magnitude.
