WiMi Releases Next-Generation Hybrid Quantum Neural Network Structure Technology, Breaking Through the Bottleneck of Image Multi-Classification

Rhea-AI Summary

WiMi (NASDAQ: WIMI) announced a next‑generation hybrid quantum neural network (H-QNN) for image multi‑classification on Dec 22, 2025. The H-QNN combines classical convolutional neural networks for spatial feature extraction with quantum neural networks for high‑dimensional nonlinear mapping, using a three‑stage design: feature dimensionality reduction & encoding, quantum state transformation, and hybrid decision & transfer learning. Key technical elements include PCA plus angle encoding, parameter sharing to mitigate barren plateaus, an early stopping strategy using quantum Fidelity, and an FPGA‑accelerated quantum simulation module claiming nanosecond‑level state updates and superior training speed versus pure CPU/GPU simulations.

The design supports simulation and hardware QPU execution and emphasizes transfer learning to reduce epochs and improve stability in multi‑class tasks.

Positive

• Three‑stage H‑QNN architecture: convolutional–quantum–hybrid
• Uses PCA + angle encoding to address quantum encoding dimensionality limits
• FPGA module enables quantum state updates at nanosecond‑level response times

Negative

• Claims of performance 'far exceeding' CPU/GPU lack disclosed numeric benchmarks
• Practical QPU deployment and real‑world inference performance are not quantified

Key Figures

Main modules 3 modules Feature reduction/encoding, quantum state transformation, hybrid decision/transfer

Architecture stages 3 stages Convolutional feature extraction–quantum mapping–hybrid decision-making structure

Market Reality Check

$2.88 Last Close

Volume Volume 65,764 is 5% above the 20-day average of 62,907, showing only modest interest pre-news. normal

Technical Shares at $2.92 are trading below the 200-day MA $4.18 and sit 90% under the 52-week high $29.20, though 30.64% above the 52-week low $2.2351.

Peers on Argus 1 Up 1 Down

Within Communication Services/Advertising peers, moves were mixed: ABLV up 3.8%, SWAG up 2.86%, ACCS and MCTR modestly positive, FLNT flat. Momentum scanner shows KRKR up 8.5% while ABLV down 6.83%, underscoring stock-specific rather than broad sector AI/quantum trading.

Historical Context

Date	Event	Sentiment	Move	Catalyst
Dec 04	Quantum-classical architecture	Positive	+1.9%	Hybrid quantum-classical learning design for multi-class image classification.
Dec 04	Quantum-classical architecture	Positive	+1.9%	Repeat release on hybrid quantum-classical image classification model.
Nov 20	Quantum GAN research	Positive	-5.6%	Dual-discriminator QGAN architecture targeting more stable and diverse image generation.
Nov 14	QGAN & hybrid model	Positive	-1.7%	Quantum GAN and hybrid CNN model for faster, more accurate synthetic image tasks.
Oct 24	Post-quantum blockchain	Positive	-0.5%	Blockchain privacy system using post-quantum cryptography and threshold key sharing.

Pattern Detected

Quantum/AI research announcements often received muted or negative next-day reactions, with only some events seeing modest gains despite consistently positive R&D framing.

Recent Company History

Over the past months, WiMi repeatedly highlighted quantum and AI R&D initiatives. On Oct 24, a post‑quantum blockchain privacy system saw a -0.51% move. Quantum GAN and hybrid models announced on Nov 14 and Nov 20 led to -1.68% and -5.61% moves, respectively. A hybrid quantum-classical architecture on Dec 4 coincided with a 1.95% gain. Today’s hybrid quantum neural network update continues this technology-focused trajectory in intelligent vision.

Market Pulse Summary

This announcement details a hybrid quantum neural network (H-QNN) for image multi-classification, combining CNN-based feature extraction with quantum circuits and transfer learning to enhance accuracy and efficiency. It follows earlier quantum AI work such as SHQCNN, SQ-QNN, and QDCNN, indicating a sustained R&D focus. Investors may monitor how these architectures progress from experimental setups toward concrete deployments in intelligent vision and related applications, and whether future filings clarify commercial impact.

Key Terms

convolutional neural networks technical

"integrates the spatial feature extraction capabilities of classical convolutional neural networks (CNN)"

Convolutional neural networks are a type of computer program that learns to recognize patterns in images, charts, or other grid-like data by automatically finding and combining simple features into more complex ones, much like how the eye and brain detect edges and shapes. Investors care because these models power tools for automated image analysis, fraud detection, algorithmic trading signals and product improvements; their use can change costs, competitive advantage and the accuracy of data-driven decisions.

fpga technical

"quantum modules are executed in quantum simulators or FPGA-accelerated quantum kernel"

A field-programmable gate array (FPGA) is a type of computer chip whose internal wiring can be changed after it is made, allowing engineers to program custom hardware functions without designing a new chip. For investors, FPGAs matter because that flexibility lets companies quickly adapt products to new software, standards, or customer needs—like a toolbox that can be rearranged to build different machines—so demand and pricing can shift with trends in data centers, telecommunications, AI, and specialized electronics.

AI-generated analysis. Not financial advice.

BEIJING, Dec. 22, 2025 /PRNewswire/ -- WiMi Hologram Cloud Inc. (NASDAQ: WiMi) ("WiMi" or the "Company"), a leading global Hologram Augmented Reality ("AR") Technology provider, launched a hybrid quantum neural network structure (H-QNN) for image multi-classification. This technology organically integrates the spatial feature extraction capabilities of classical convolutional neural networks (CNN) with the high-dimensional nonlinear mapping features of quantum neural networks (QNN), forming a new type of hybrid structure that possesses stronger generalization ability and computational efficiency in multi-class classification scenarios. This technology not only systematically optimizes the quantum-classical hybrid learning system in theory but also achieves classification accuracy and stability superior to similar algorithms in actual experiments, laying a solid technical foundation for quantum intelligent vision systems.

The design of this hybrid quantum neural network (H-QNN) follows the principle of classical responsible for abstraction and quantum responsible for discrimination. The overall system consists of three main modules: feature dimensionality reduction and encoding module, quantum state transformation module, and hybrid decision and transfer learning module.

First, the feature dimensionality reduction and encoding module is based on the classical convolutional neural network (CNN) structure, extracting low-dimensional feature representations of images through several convolutional layers and pooling layers. The feature vectors after PCA dimensionality reduction are standardized and then input into the quantum encoding circuit. At this stage, WiMi adopts an improved angle encoding method (Angle Embedding) to map real-valued features to quantum state amplitudes, and achieves efficient encoding through multi-layer quantum rotation gates (Ry, Rz), thereby reducing quantum gate depth and lowering encoding noise.

Next, the quantum state transformation module undertakes the core tasks of high-dimensional feature mapping and nonlinear discrimination. This module includes several layers of quantum circuits, with each layer composed of parameterized rotation gates and controlled entanglement gates (CNOT or CZ), forming nonlinear coupling and entanglement of quantum states. To alleviate gradient vanishing, WiMi adopts a reconfigurable parameter sharing strategy, allowing different quantum layers to share some trainable parameters, while introducing mixed state perturbations to maintain gradient balance during the training process. This structural design effectively avoids the barren plateau phenomenon, enabling the model to maintain stable convergence in multi-class tasks.

Finally, the hybrid decision and transfer learning module integrates the results of quantum computing with the classical decision layer. The measurement probability distribution output by the quantum circuit is converted into feature vectors and fused with the output of the classical fully connected layer. This fused vector is input into the Softmax layer for final classification judgment. To further enhance the generalization performance in multi-class tasks, WiMi introduces a transfer learning mechanism, migrating the parameters of quantum layers pre-trained in small-sample tasks to new tasks, thereby reducing the number of training epochs and enhancing model stability.

In actual implementation, this structure supports running on simulation environments and hardware quantum processing units (QPU). The simulation environment uses high-performance GPU clusters to complete training of classical modules, while quantum modules are executed in quantum simulators or FPGA-accelerated quantum kernel estimation environments, achieving heterogeneous collaboration of classical and quantum computing resources.

The core innovation points of this technology are mainly embodied in the following aspects.

First, at the architectural design level, it achieves deep integration of convolutional neural networks (CNN) and quantum neural networks (QNN). Traditional quantum hybrid models usually simply embed the quantum part as a classification head, whereas the H-QNN proposed in this research adopts a three-stage distributed structure of "convolutional feature extraction—quantum mapping—hybrid decision-making", enabling the quantum part not only to undertake nonlinear discrimination but also to achieve information reconstruction at the feature space level.

Second, at the encoding strategy level, the joint dimensionality reduction scheme of angle encoding and principal component analysis (PCA) proposed by WiMi effectively solves the quantum encoding dimension limitation problem. By optimizing the cumulative variance contribution rate of PCA, it ensures that the mapping between input features and quantum state amplitudes maintains high information fidelity, thereby maximizing the utilization rate of quantum information.

Third, at the training strategy level, WiMi introduces a transfer learning mechanism and parameter sharing structure. Traditional quantum neural networks often face risks of gradient vanishing and overfitting in multi-class classification training, while parameter sharing can establish balanced gradient flow between different quantum layers, and the transfer learning mechanism enables the model to achieve rapid convergence on new tasks with fewer training epochs. In addition, WiMi designs an early stopping strategy based on the quantum Fidelity metric, which determines whether the training has reached the optimal point by monitoring the stability of quantum state evolution, thereby preventing overfitting.

Finally, at the system implementation level, it adopts a heterogeneous computing architecture, running the classical computing part on CPU/GPU platforms, while the quantum part is executed in quantum simulation modules implemented on FPGA. The FPGA module realizes reconfigurable execution logic for parameterized quantum circuits, capable of completing quantum state updates within nanosecond-level response times, thereby significantly improving the overall training speed of the system. This hybrid computing architecture demonstrates performance advantages far exceeding pure CPU or GPU simulations in experiments.

The proposal of WiMi's hybrid quantum neural network structure marks a key step in quantum artificial intelligence research moving from theoretical exploration toward practical applications. It not only demonstrates the potential advantages of quantum computing in the field of machine learning but also provides an engineered compromise solution for the current performance bottlenecks of quantum hardware. By embedding trainable quantum layers into the foundation of classical neural networks, this technology achieves efficient utilization of quantum computing resources, enabling quantum advantages to be embodied in real visual tasks. In the future, quantum intelligence will no longer be merely a theoretical conception but will deeply integrate with fields such as deep learning, computer vision, and edge computing, becoming an important driving force for promoting the development of intelligent society. Let quantum intelligence move from the laboratory to the real world, and let quantum technology truly serve industrial upgrades and the expansion of human cognition.

About WiMi Hologram Cloud

WiMi Hologram Cloud Inc. (NASDAQ: WiMi) focuses on holographic cloud services, primarily concentrating on professional fields such as in-vehicle AR holographic HUD, 3D holographic pulse LiDAR, head-mounted light field holographic devices, holographic semiconductors, holographic cloud software, holographic car navigation, metaverse holographic AR/VR devices, and metaverse holographic cloud software. It covers multiple aspects of holographic AR technologies, including in-vehicle holographic AR technology, 3D holographic pulse LiDAR technology, holographic vision semiconductor technology, holographic software development, holographic AR virtual advertising technology, holographic AR virtual entertainment technology, holographic ARSDK payment, interactive holographic virtual communication, metaverse holographic AR technology, and metaverse virtual cloud services. WiMi is a comprehensive holographic cloud technology solution provider. For more information, please visit http://ir.wimiar.com.

Translation Disclaimer

The original version of this announcement is the officially authorized and only legally binding version. If there are any inconsistencies or differences in meaning between the Chinese translation and the original version, the original version shall prevail. WiMi Hologram Cloud Inc. and related institutions and individuals make no guarantees regarding the translated version and assume no responsibility for any direct or indirect losses caused by translation inaccuracies.

Investor Inquiries, please contact:

WIMI Hologram Cloud Inc.
Email: pr@wimiar.com

ICR, LLC
Robin Yang
Tel: +1 (646) 975-9495
Email: wimi@icrinc.com

View original content:https://www.prnewswire.com/news-releases/wimi-releases-next-generation-hybrid-quantum-neural-network-structure-technology-breaking-through-the-bottleneck-of-image-multi-classification-302648081.html

SOURCE WiMi Hologram Cloud Inc.

FAQ

What is WiMi's H‑QNN announced on Dec 22, 2025 (WIMI)?

WiMi's H‑QNN is a hybrid model combining CNN feature extraction with quantum neural network mapping for image multi‑classification.

How does the H‑QNN encode image features for quantum processing in WIMI's design?

Features are reduced by PCA, standardized, then mapped into quantum states using an improved angle encoding with multi‑layer Ry/Rz rotations.

What training strategies does WIMI use to avoid barren plateaus in the H‑QNN?

WIMI applies parameter sharing across quantum layers, mixed state perturbations, and an early stopping rule based on a quantum Fidelity metric.

Can WIMI's H‑QNN run on real quantum hardware (WIMI)?

The system supports both quantum simulators and hardware QPU execution; quantum modules may run in simulators or FPGA‑accelerated environments.

What performance gains does WiMi claim for the H‑QNN (WIMI)?

WiMi reports classification accuracy and stability superior to similar algorithms and says the heterogeneous architecture shows speed advantages over pure CPU/GPU simulations.

Does WiMi provide commercial timelines or numeric benchmarks for the H‑QNN (WIMI)?

No commercial deployment dates or detailed numeric benchmark results are disclosed in the announcement.