WiMi Hologram Cloud Inc. Develops Single-Qudit Quantum Neural Network Technology

WiMi Hologram Cloud Inc. has announced the development of single-qubit quantum neural network technology, a breakthrough in high-dimensional quantum systems for efficient learning. This technology paves the way for the deep integration of future quantum computing and artificial intelligence. The current reliance on billions of parameters and massive data center resources in traditional neural networks has become a bottleneck in AI development, with significant increases in power consumption and hardware costs. WiMi's innovation promises to break through these resource limitations and improve model compactness and computational efficiency.

Key Takeaways:

• WiMi's single-qudit quantum neural network technology utilizes superposition and entanglement to achieve natural representation of high-dimensional data spaces, breaking the limitations of classical computing.
• The technology directly handles multi-class classification tasks through the state space of a single high-dimensional qudit, unlike classical neural networks that rely on thousands of neurons and complex hierarchical structures.
• SQ-QNN leverages the high-dimensional characteristics of quantum systems to efficiently encode and distinguish category information within a compact circuit scale.
• The technology uses the Cayley transform of skew-symmetric matrices to construct unitary operators, which ensures mathematical stability and efficiency in quantum circuit implementation.
• QiMi proposes a hybrid training method that combines extended activation functions with Support Vector Machine (SVM) optimization, ensuring the stability of parameter optimization and global optimal solutions.
• The SQ-QNN technology reduces circuit depth, training overhead, and the burden on quantum hardware, improving training efficiency and convergence to the global optimal solution.
• WiMi's single-qudit quantum neural network technology is a significant step towards promoting industrial progress in quantum machine learning.

Statistics:

• $d$-dimensional qudit system is constructed to carry data in multi-class classification problems.
• Cayley transform of skew-symmetric matrices generates $d$-dimensional unitary operators preserving unitarity and physical rationality.
• Hybrid quantum-classical training method combines extended activation functions with Support Vector Machine (SVM) optimization, providing efficient parameter search and network convergence.
• The technology reduces circuit depth by achieving complex decision boundaries through single-step evolution, and it introduces a compact circuit scale for efficient encoding and distinguish category information.

Sources:

• WiMi Hologram Cloud Inc. press release dated October 20, 2025
• Original publication date: October 20, 2025
• Globe Newswire release