WiMi Has Developed a Scalable Quantum Neural Network (SQNN) Technology Based on Multi-quantum-device Collaborative Computing
BEIJING, Sept. 19, 2025 /PRNewswire/ -- WiMi Hologram Cloud Inc. (NASDAQ: WiMi) ("WiMi" or the "Company"), a leading global Hologram Augmented Reality ("AR") Technology provider, today announced the development of a Scalable Quantum Neural Network (SQNN) technology based on multi-quantum-device collaborative computing. This technology utilizes multiple small quantum devices as quantum feature extractors, which extract local features from input data in parallel. The extracted local features are then aggregated into a quantum predictor through classical communication channels to accomplish the final classification task.
This technology aims to overcome the limitations of current quantum computing hardware by enabling multiple small quantum devices to work collaboratively, thereby building an efficient and scalable quantum neural network system. The technology not only achieves classification accuracy comparable to traditional Quantum Neural Networks (QNN) in theory but also introduces a novel approach to optimizing the utilization of quantum computing resources and enhancing data efficiency.
The core architecture of WiMi's SQNN system consists of three main components:
Quantum Feature Extractor: The quantum feature extractor is responsible for extracting local features from input data. Each quantum device can independently perform feature extraction tasks using Variational Quantum Circuits (VQC) to encode and transform input data. Since these devices operate independently, they can flexibly adapt to quantum devices of different sizes. For instance, larger quantum devices can handle more complex data patterns, while smaller quantum devices can process simpler local features.
Classical Communication Channel: In the SQNN framework, quantum feature extractors transmit the extracted local features to a central computing node via a classical communication channel. This communication process is similar to the concept of Federated Learning, where different computing units process data independently, but the final decision-making process relies on the integration of global information.
Quantum Predictor: The quantum predictor serves as the core computational unit of the entire SQNN system. It receives feature information from multiple quantum feature extractors and performs the final classification decision using quantum circuits. The quantum predictor can employ more complex quantum circuits to optimize classification accuracy and dynamically adjust its computational approach based on the scale of the data.
The technical implementation of WiMi's SQNN involves the following steps: Frist, data Preprocessing and Quantum Encoding: Before entering the quantum system, input data undergoes classical preprocessing operations such as standardization and dimensionality reduction. The data is then mapped to quantum states using encoding methods such as Amplitude Encoding or Angle Encoding. Then, sub-feature Extraction: Each quantum device performs independent feature extraction tasks using Parameterized Quantum Circuits (PQC) to transform features and generate local feature representations. Besides, feature Aggregation and Classification: The output of quantum feature extractors is transmitted to a central node via a classical communication channel. The quantum predictor then aggregates the features and performs the final classification task. Finally, parameter Optimization and Training: SQNN employs Variational Quantum Optimization for training. A classical optimizer, such as gradient descent, is used to adjust the quantum circuit parameters to minimize classification error.
Compared to traditional QNNs, WiMi's SQNN offers the following significant advantages:
Improved Data Utilization: Since SQNN leverages multiple quantum devices for collaborative computing, it can utilize data more efficiently without compromising data integrity due to the qubit limitations of a single device.
Enhanced Computational Scale: By coordinating multiple small quantum devices, SQNN can handle larger-scale computational tasks without relying on a single high-performance quantum computer. This modular approach also makes SQNN more scalable.
Optimized Computing Resources: SQNN allows different types of quantum devices to work together, enabling more flexible resource allocation. For example, when the workload is small, only a subset of quantum feature extractors can be activated, whereas for large-scale computing tasks, more quantum devices can be utilized to improve computational efficiency.
Experiments conducted on multiple benchmark datasets demonstrate that WiMi's SQNN achieves classification accuracy comparable to traditional QNNs at the same scale. Additionally, since SQNN utilizes multiple quantum devices for parallel computing, its training efficiency is significantly improved compared to QNNs that rely on a single quantum device.
Furthermore, experimental results show that as the number of participating quantum devices increases, both the classification accuracy and computation speed of SQNN improve significantly. This indicates that the approach has strong scalability as quantum computing hardware continues to evolve.
Despite the promising experimental results achieved under current hardware conditions, several key challenges remain to be addressed. For example, optimizing the interconnection of quantum devices to enhance efficiency while minimizing communication costs, and further refining SQNN's quantum circuit design to reduce noise interference and improve computational accuracy.
WiMi's Scalable Quantum Neural Network (SQNN) provides an innovative solution for quantum machine learning, enabling multiple small quantum devices to work collaboratively for efficient classification tasks. Experimental results indicate that SQNN offers strong computational performance and scalability, laying a solid foundation for the integration of quantum computing and artificial intelligence. As quantum hardware continues to advance, SQNN is expected to become a crucial component of large-scale quantum machine learning systems, driving revolutionary changes in AI and data science.
About WiMi Hologram Cloud
WiMi Hologram Cloud, Inc. (NASDAQ:WiMi) is a holographic cloud comprehensive technical solution provider that focuses on professional areas including holographic AR automotive HUD software, 3D holographic pulse LiDAR, head-mounted light field holographic equipment, holographic semiconductor, holographic cloud software, holographic car navigation and others. Its services and holographic AR technologies include holographic AR automotive application, 3D holographic pulse LiDAR technology, holographic vision semiconductor technology, holographic software development, holographic AR advertising technology, holographic AR entertainment technology, holographic ARSDK payment, interactive holographic communication and other holographic AR technologies.
AI
