BEIJING, Aug. 4, 2026 /PRNewswire/ -- WiMi Hologram Cloud Inc. (NASDAQ:WIMI) (&#34;WiMi&#34; or the &#34;Company&#34;), a leading global Hologram Augmented Reality (&#34;AR&#34;) Technology provider, is exploring a

<p xmlns="http://www.w3.org/1999/xhtml"><span xmlns="http://www.w3.org/1999/xhtml" class="legendSpanClass">BEIJING</span>, <span xmlns="http://www.w3.org/1999/xhtml" class="legendSpanClass">Aug. 4, 2026</span> /PRNewswire/ -- WiMi Hologram Cloud Inc. (NASDAQ:<a class="ticker" href="https://www.benzinga.com/quote/WIMI">WIMI</a>) (&#34;WiMi&#34; or the &#34;Company&#34;), a leading global Hologram Augmented Reality (&#34;AR&#34;) Technology provider, is exploring a federated training framework for hybrid quantum-classical machine learning models, deeply integrating quantum neural networks (QNN) with classical pre-trained convolutional models, achieving dual breakthroughs in model accuracy and training efficiency through a distributed quantum computing architecture, and opening up innovative technical paths for data privacy protection and large-scale model training.</p>
<p xmlns="http://www.w3.org/1999/xhtml">Quantum computers, through the superposition and entanglement states of qubits, can process an exponential number of computational tasks in parallel. This characteristic gives quantum machine learning (QML) potential advantages in tasks such as optimization problems and feature extraction. However, current quantum hardware is limited by factors such as noise and decoherence time, making it difficult to directly support the training of large-scale deep learning models. Federated learning uses a distributed training mechanism of &#34;data stationary, model moving&#34; to allow local nodes to share only model parameters to a central server for aggregation while retaining the original data. This mode effectively avoids the leakage risks brought by centralized data storage, but traditional FL frameworks still face two major challenges: first, the computational overhead of classical neural networks on edge devices leads to low training efficiency; second, the communication costs in the model parameter transmission process surge with the growth of node scale.</p><p xmlns="http://www.w3.org/1999/xhtml">The fusion of quantum computing and federated learning provides solutions to the above problems. The parallel computing capability of quantum algorithms can reduce the complexity of optimization problems in high-dimensional feature spaces from exponential to polynomial levels, while the distributed architecture of federated learning provides a natural carrier for the engineering deployment of quantum models. The deeply collaborative hybrid architecture constructed by WiMi uses classical pre-trained convolutional models to handle underlying feature extraction tasks, leveraging their mature advantages in learning basic features such as image textures and edges, with quantum neural networks composed of parameterized quantum circuits (PQC) responsible for nonlinear mapping of higher-level abstract features, enhancing feature expression capabilities through quantum entanglement characteristics; ultimately, multi-node model collaborative optimization is achieved through the federated learning framework.</p><p xmlns="http://www.w3.org/1999/xhtml">This innovative architecture demonstrates three major breakthroughs in its technical implementation. At the model design level, it adopts a hybrid quantum-classical convolutional neural network (SHQCNN) structure, where the input layer maps image data from low-dimensional space to high-dimensional quantum feature space through kernel encoding methods, solving the adaptation challenges between quantum states and classical data. The hidden layer innovatively introduces enhanced variational quantum circuits, which, through optimization of quantum gate combinations and adaptive parameter adjustment strategies, achieve feature extraction capabilities comparable to deep classical CNNs with only dozens of qubits, while avoiding the accumulation of quantum noise as circuit depth increases. The output layer employs the mini-batch gradient descent algorithm, accelerating model convergence through a high-frequency weight update mechanism, making the fusion decision of quantum-classical features more efficient. In the federated training mechanism, WiMi designs a layered aggregation communication protocol. Local nodes retain only the classical pre-trained convolutional base model and lightweight quantum processors; after completing local data feature encoding and model training through quantum circuits, only the encrypted quantum parameter gradients are uploaded to the central server, rather than raw feature data. The central node uses quantum state technology to complete cross-node parameter aggregation, then generates global update parameters through a classical optimizer and distributes them to each node.</p><p xmlns="http://www.w3.org/1999/xhtml">The hybrid quantum-classical federated learning framework researched by WiMi marks a paradigm shift in machine learning from classical computing to quantum-enhanced computing, solving the core issues of data privacy and training efficiency. With the maturation of quantum hardware and optimization of algorithms, this innovative technology is expected to lead the transformation of next-generation artificial intelligence infrastructure.</p><p xmlns="http://www.w3.org/1999/xhtml">About WiMi Hologram Cloud</p><p xmlns="http://www.w3.org/1999/xhtml">WiMi Hologram Cloud Inc. (NASDAQ:<a class="ticker" href="https://www.benzinga.com/quote/WIMI">WIMI</a>) 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 <a xmlns="http://www.w3.org/1999/xhtml" href="http://ir.wimiar.com/" target="_blank" rel="nofollow">http://ir.wimiar.com</a>.</p><p xmlns="http://www.w3.org/1999/xhtml">Translation Disclaimer</p><p xmlns="http://www.w3.org/1999/xhtml">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.</p><p xmlns="http://www.w3.org/1999/xhtml"> </p>



<p xmlns="http://www.w3.org/1999/xhtml" id="PURL"><img xmlns="http://www.w3.org/1999/xhtml" title="Cision" width="12" height="12" alt="Cision" src="https://edge.prnewswire.com/c/img/favicon.png?sn=CN18789&amp;sd=2026-08-04"></img> View original content:<a xmlns="http://www.w3.org/1999/xhtml" id="PRNURL" rel="nofollow" href="https://www.prnewswire.com/news-releases/wimi-explores-federated-training-framework-for-hybrid-quantum-classical-machine-learning-models-302842447.html" target="_blank">https://www.prnewswire.com/news-releases/wimi-explores-federated-training-framework-for-hybrid-quantum-classical-machine-learning-models-302842447.html</a></p><p xmlns="http://www.w3.org/1999/xhtml">SOURCE WiMi Hologram Cloud Inc.</p>

WIMI

WiMi Explores Federated Training Framework for Hybrid Quantum-Classical Machine Learning Models

WiMi Advances Quantum-Classical Federated ML Framework for Long-Term AI Advantage

<section class="insight" data-version="1">
  <section data-block="information">
    <h3>Information</h3>
    <p>BEIJING, Aug. 4, 2026 /PRNewswire/ -- WiMi Hologram Cloud Inc. (NASDAQ:WIMI) (&amp;#34;WiMi&amp;#34; or the &amp;#34;Company&amp;#34;), a leading global Hologram Augmented Reality (&amp;#34;AR&amp;#34;) Technology provider, is exploring a</p>
  </section>

  <section data-block="ai-summary-paragraph">
    <h3>AI Summary</h3>
    <p>WiMi Hologram Cloud disclosed exploration of a federated training framework for hybrid quantum-classical ML, integrating quantum neural networks with classical CNNs to boost accuracy and training efficiency. The SHQCNN design uses dozens of qubits and encrypted gradient sharing across nodes, addressing data privacy and scalability. While promising, this remains early-stage with no near-term revenue impact, so milestone progress will drive any valuation upside.</p>
  </section>

  <section data-block="ai-summary-bullets">
    <h3>Key Points</h3>
    <ul><li>WiMi explores a federated hybrid quantum-classical ML framework.</li><li>Hybrid SHQCNN uses classical CNN base with PQC quantum features.</li><li>Federated setup uploads only encrypted gradients, not raw data.</li><li>Data privacy and training efficiency could transform AI infrastructure.</li><li>No immediate revenue impact; milestones needed for material valuation.</li></ul>
  </section>

  <section data-block="companies-mentioned">
    <h3>Companies Mentioned</h3>
    <ul><li><strong>WiMi Hologram Cloud Inc. (WIMI)</strong>: Subject of the release; potential long-term AI/quantum moat; no near-term revenue disclosed.</li><li><strong>N/A (N/A)</strong>: No other companies named in the release.</li></ul>
  </section>

  <section data-block="category">
    <h3>Industry News</h3>
    <p>Category: Industry News. The article centers on WiMi&#039;s R&amp;D initiative and potential tech moat rather than earnings, M&amp;A, or promotions, making it industry-relevant but not an immediate fundamental driver.</p>
  </section>

  <section data-block="market-moving">
    <h3>Market-Moving</h3>
    <ul><li>Quantum-AI milestones at WiMi could attract analyst interest and affect valuation.</li><li>Lack of revenue data limits upside potential and may cap near-term moves.</li><li>Any partnerships or pilots with hardware providers could trigger volatility.</li></ul>
  </section>

  <section data-block="key-facts">
    <h3>Key Facts</h3>
    <ul><li>Date: Aug. 4, 2026.</li><li>Company: WiMi Hologram Cloud Inc., NASDAQ: WIMI.</li><li>Focus: federated training framework for hybrid quantum-classical ML models.</li><li>Architecture: SHQCNN with kernel encoding and PQC; encrypted gradient aggregation.</li><li>Technical claim: high-dimensional feature mapping with dozens of qubits; polynomial complexity.</li><li>Data handling: only encrypted gradients uploaded; raw data remains local.</li></ul>
  </section>

  <section data-block="trading-thesis">
    <h3>Trading Thesis</h3>
    <p>Over 12–24 months, WIMI could trend higher on quantum-AI milestones, absent near-term revenue catalysts.</p>
  </section>
</section>


<p xmlns="http://www.w3.org/1999/xhtml"><span xmlns="http://www.w3.org/1999/xhtml" class="legendSpanClass">BEIJING</span>, <span xmlns="http://www.w3.org/1999/xhtml" class="legendSpanClass">Aug. 4, 2026</span> /PRNewswire/ -- WiMi Hologram Cloud Inc. (NASDAQ:<a class="ticker" href="https://www.benzinga.com/quote/WIMI">WIMI</a>) (&#34;WiMi&#34; or the &#34;Company&#34;), a leading global Hologram Augmented Reality (&#34;AR&#34;) Technology provider, is exploring a federated training framework for hybrid quantum-classical machine learning models, deeply integrating quantum neural networks (QNN) with classical pre-trained convolutional models, achieving dual breakthroughs in model accuracy and training efficiency through a distributed quantum computing architecture, and opening up innovative technical paths for data privacy protection and large-scale model training.</p>
<p xmlns="http://www.w3.org/1999/xhtml">Quantum computers, through the superposition and entanglement states of qubits, can process an exponential number of computational tasks in parallel. This characteristic gives quantum machine learning (QML) potential advantages in tasks such as optimization problems and feature extraction. However, current quantum hardware is limited by factors such as noise and decoherence time, making it difficult to directly support the training of large-scale deep learning models. Federated learning uses a distributed training mechanism of &#34;data stationary, model moving&#34; to allow local nodes to share only model parameters to a central server for aggregation while retaining the original data. This mode effectively avoids the leakage risks brought by centralized data storage, but traditional FL frameworks still face two major challenges: first, the computational overhead of classical neural networks on edge devices leads to low training efficiency; second, the communication costs in the model parameter transmission process surge with the growth of node scale.</p><p xmlns="http://www.w3.org/1999/xhtml">The fusion of quantum computing and federated learning provides solutions to the above problems. The parallel computing capability of quantum algorithms can reduce the complexity of optimization problems in high-dimensional feature spaces from exponential to polynomial levels, while the distributed architecture of federated learning provides a natural carrier for the engineering deployment of quantum models. The deeply collaborative hybrid architecture constructed by WiMi uses classical pre-trained convolutional models to handle underlying feature extraction tasks, leveraging their mature advantages in learning basic features such as image textures and edges, with quantum neural networks composed of parameterized quantum circuits (PQC) responsible for nonlinear mapping of higher-level abstract features, enhancing feature expression capabilities through quantum entanglement characteristics; ultimately, multi-node model collaborative optimization is achieved through the federated learning framework.</p><p xmlns="http://www.w3.org/1999/xhtml">This innovative architecture demonstrates three major breakthroughs in its technical implementation. At the model design level, it adopts a hybrid quantum-classical convolutional neural network (SHQCNN) structure, where the input layer maps image data from low-dimensional space to high-dimensional quantum feature space through kernel encoding methods, solving the adaptation challenges between quantum states and classical data. The hidden layer innovatively introduces enhanced variational quantum circuits, which, through optimization of quantum gate combinations and adaptive parameter adjustment strategies, achieve feature extraction capabilities comparable to deep classical CNNs with only dozens of qubits, while avoiding the accumulation of quantum noise as circuit depth increases. The output layer employs the mini-batch gradient descent algorithm, accelerating model convergence through a high-frequency weight update mechanism, making the fusion decision of quantum-classical features more efficient. In the federated training mechanism, WiMi designs a layered aggregation communication protocol. Local nodes retain only the classical pre-trained convolutional base model and lightweight quantum processors; after completing local data feature encoding and model training through quantum circuits, only the encrypted quantum parameter gradients are uploaded to the central server, rather than raw feature data. The central node uses quantum state technology to complete cross-node parameter aggregation, then generates global update parameters through a classical optimizer and distributes them to each node.</p><p xmlns="http://www.w3.org/1999/xhtml">The hybrid quantum-classical federated learning framework researched by WiMi marks a paradigm shift in machine learning from classical computing to quantum-enhanced computing, solving the core issues of data privacy and training efficiency. With the maturation of quantum hardware and optimization of algorithms, this innovative technology is expected to lead the transformation of next-generation artificial intelligence infrastructure.</p><p xmlns="http://www.w3.org/1999/xhtml">About WiMi Hologram Cloud</p><p xmlns="http://www.w3.org/1999/xhtml">WiMi Hologram Cloud Inc. (NASDAQ:<a class="ticker" href="https://www.benzinga.com/quote/WIMI">WIMI</a>) 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 <a xmlns="http://www.w3.org/1999/xhtml" href="http://ir.wimiar.com/" target="_blank" rel="nofollow">http://ir.wimiar.com</a>.</p><p xmlns="http://www.w3.org/1999/xhtml">Translation Disclaimer</p><p xmlns="http://www.w3.org/1999/xhtml">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.</p><p xmlns="http://www.w3.org/1999/xhtml"> </p>



<p xmlns="http://www.w3.org/1999/xhtml" id="PURL"><img xmlns="http://www.w3.org/1999/xhtml" title="Cision" width="12" height="12" alt="Cision" src="https://edge.prnewswire.com/c/img/favicon.png?sn=CN18789&amp;sd=2026-08-04"></img> View original content:<a xmlns="http://www.w3.org/1999/xhtml" id="PRNURL" rel="nofollow" href="https://www.prnewswire.com/news-releases/wimi-explores-federated-training-framework-for-hybrid-quantum-classical-machine-learning-models-302842447.html" target="_blank">https://www.prnewswire.com/news-releases/wimi-explores-federated-training-framework-for-hybrid-quantum-classical-machine-learning-models-302842447.html</a></p><p xmlns="http://www.w3.org/1999/xhtml">SOURCE WiMi Hologram Cloud Inc.</p>

WiMi Hologram Cloud disclosed exploration of a federated training framework for hybrid quantum-classical ML, integrating quantum neural networks with classical CNNs to boost accuracy and training efficiency. The SHQCNN design uses dozens of qubits and encrypted gradient sharing across nodes, addressing data privacy and scalability. While promising, this remains early-stage with no near-term revenue impact, so milestone progress will drive any valuation upside.

News insight

WiMi Explores Federated Training Framework for Hybrid Quantum-Classical Machine Learning Models

AI summary

Sentiment rationale

Key facts

Industry News