Distributed Quantum Neural Networks on Distributed Photonic Quantum Computing

2025-05-13

AI Summary

To address the inefficiency of parameter training and poor classical-quantum integration in photonic quantum neural networks (QNNs) for distributed photonic quantum computing, this paper proposes a photonic quantum-classical hybrid training framework. It leverages photonic QNNs to generate high-dimensional probability distributions, which are then compressed into compact classical weights via matrix product states (MPS), enabling end-to-end differentiable training. Crucially, quantum hardware is unnecessary during inference, preserving quantum representational power while ensuring classical deployability. The method integrates universal linear optical interferometer decomposition, photon-counting statistical modeling, and noise-robust simulation. On MNIST, it achieves 95.50% accuracy with only 3,292 parameters—over 2× more parameter-efficient than comparable classical baselines, with <3% accuracy degradation. Ablation studies and noise-resilient simulations validate both feasibility and the essential role of quantum components.

Technology Category

• Machine Learning: Quantum Machine Learning
• Computer Vision: Learning & Optimization for CV
• Search and Optimization: Learning to Search

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

We introduce a distributed quantum-classical framework that synergizes photonic quantum neural networks (QNNs) with matrix-product-state (MPS) mapping to achieve parameter-efficient training of classical neural networks. By leveraging universal linear-optical decompositions of $M$-mode interferometers and photon-counting measurement statistics, our architecture generates neural parameters through a hybrid quantum-classical workflow: photonic QNNs with $M(M+1)/2$ trainable parameters produce high-dimensional probability distributions that are mapped to classical network weights via an MPS model with bond dimension $chi$. Empirical validation on MNIST classification demonstrates that photonic QT achieves an accuracy of $95.50% pm 0.84%$ using 3,292 parameters ($chi = 10$), compared to $96.89% pm 0.31%$ for classical baselines with 6,690 parameters. Moreover, a ten-fold compression ratio is achieved at $chi = 4$, with a relative accuracy loss of less than $3%$. The framework outperforms classical compression techniques (weight sharing/pruning) by 6--12% absolute accuracy while eliminating quantum hardware requirements during inference through classical deployment of compressed parameters. Simulations incorporating realistic photonic noise demonstrate the framework's robustness to near-term hardware imperfections. Ablation studies confirm quantum necessity: replacing photonic QNNs with random inputs collapses accuracy to chance level ($10.0% pm 0.5%$). Photonic quantum computing's room-temperature operation, inherent scalability through spatial-mode multiplexing, and HPC-integrated architecture establish a practical pathway for distributed quantum machine learning, combining the expressivity of photonic Hilbert spaces with the deployability of classical neural networks.

Problem

• Achieving parameter-efficient training of classical neural networks using photonic QNNs and MPS mapping.
• Demonstrating quantum advantage in neural network compression and accuracy over classical methods.
• Validating robustness of distributed quantum-classical framework to realistic photonic noise.

Innovation

• Distributed quantum-classical framework with photonic QNNs
• Matrix-product-state mapping for parameter-efficient training
• Hybrid quantum-classical workflow for neural parameters