How Quantum Computing Is Powering the Next Wave of AI

Quantum Computing Basics

Quantum computing uses qubits that can exist in superposition of 0 and 1, exploiting coherence and entanglement. This provides exponential speed‑up for certain problems compared with classical bits that are deterministic.

Example: distinguishing barcodes. A classical algorithm queries each position sequentially; a quantum algorithm can query a superposition of positions in a single step, obtaining global information with far fewer queries.

Quantum Artificial Intelligence

Quantum AI (quantum machine learning, QML) studies the intersection of machine learning and quantum information processing. Quantum algorithms can give polynomial or exponential acceleration for learning tasks, while ML techniques help design better quantum protocols, error‑correction, and noise‑mitigation.

QianJiang (量桨) – Cloud‑Integrated Quantum Machine‑Learning Platform

QianJiang (qml.baidu.com) is a cloud‑and‑hardware integrated QML platform that connects AI developers to quantum processors. It provides model libraries for combinatorial optimization, quantum simulation, and quantum networking, and a core framework comprising quantum states, quantum neural networks (QNNs), and loss functions.

Core Functions

QianJiang supports hybrid quantum‑classical workflows:

Quantum neural networks can process classical data, act as components within classical networks, and be trained end‑to‑end.

A QNN is a parameterized quantum circuit; gate parameters are optimized so that the circuit evolves to a target quantum state, analogous to classical neural networks.

Users can construct parameterized circuits with a few lines of code, optionally add realistic noise, and deploy them to quantum processors via the Qiyifu backend.

Automated circuit‑optimization reduces gate count, improving fidelity on noisy devices.

Research Highlights

Single‑qubit native QNNs can universally approximate functions; the expressive power is equivalent to a truncated Fourier series (NeurIPS 2022).

Quantum Phase Processing (QPP) extracts phase information from n‑qubit data and enables simultaneous simulation of 2ⁿ classical functions, applicable to phase estimation, Hamiltonian simulation, entropy estimation, and machine learning.

LOCCNet optimizes quantum‑entanglement protocols with machine learning, achieving up to 20 % higher fidelity than prior methods.

First practical framework for entanglement detection and quantification on near‑term devices.

Quantum natural language processing introduces self‑attention mechanisms, allowing QNNs to handle complex textual data.

Variational Shadow Quantum Learning (VSQL) reduces parameter count while improving classification accuracy on quantum data.

Hardware Experiments

Experiments on real quantum processors have demonstrated:

Entanglement generation and entropy estimation.

Variational quantum algorithms.

Simulation of the Heisenberg model.

As hardware scales, quantum advantage is expected in battery material design, high‑temperature superconductors, and other domains.

Outlook

QianJiang enables end‑to‑end QML pipelines from algorithm design to deployment on quantum hardware via Qiyifu. Ongoing improvements in platform performance and feature set are expected to accelerate both industrial applications and fundamental quantum research, deepening integration of deep‑learning frameworks with quantum compute power.

References

Yu, Z., Yao, H., Li, M., & Wang, X. (2022). Power and limitations of single‑qubit native quantum neural networks. arXiv:2205.07848. https://arxiv.org/abs/2205.07848

Wang, X., Wang, Y., Yu, Z., & Zhang, L. (2022). Quantum Phase Processing: Transform and Extract Eigen‑Information of Quantum Systems. arXiv:2209.14278. https://doi.org/10.48550/arXiv.2209.14278

Zhao, X., Zhao, B., Wang, Z., Song, Z., & Wang, X. (2021). Practical distributed quantum information processing with LOCCNet. Npj Quantum Information 7(1), 159. https://doi.org/10.1038/s41534-021-00496-x

Li, G., Zhao, X., & Wang, X. (2022). Quantum Self‑Attention Neural Networks for Text Classification. arXiv:2205.05625. https://arxiv.org/abs/2205.05625

Li, G., Song, Z., & Wang, X. (2021). VSQL: Variational shadow quantum learning for classification. AAAI 2021. https://ojs.aaai.org/index.php/AAAI/article/view/17016

Code example

the 36th Conference on Neural Information Processing Systems (NeurIPS 2022 https://nips.cc/Conferences/2022/Schedule?showEvent=53390)
[2] Wang, X., Wang, Y., Yu, Z., & Zhang, L. (2022). Quantum Phase Processing: Transform and Extract Eigen-Information of Quantum Systems
(arXiv:2209.14278)