Quantum Circuits for Quantum Convolutions: A Quantum Convolutional Autoencoder

2025-08-30

AI Summary

This work addresses the limited feature representation capacity in quantum machine learning by proposing a quantum convolution (QConv) method based on parameterized random quantum circuits. Specifically, classical input data are encoded into quantum states, and learnable quantum convolutional operations are realized via differentiable random quantum circuits, which are integrated with a classical CNN backbone to form a hybrid quantum-classical convolutional architecture. Crucially, QConv is fully simulatable on classical hardware without requiring quantum devices. Experiments demonstrate that QConv achieves classification accuracy comparable to classical CNNs on image recognition tasks while significantly accelerating training convergence—yielding an average 23% speedup. The core contribution lies in the first systematic formulation of random quantum circuits as differentiable, CNN-embeddable quantum convolutional operators, establishing a novel paradigm for lightweight, quantum-inspired representation learning.

Technology Category

• Machine Learning: Quantum Machine Learning
• Computer Vision: Representation Learning for Vision
• Search and Optimization: Non-convex Optimization

Application Category

• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Security and Privacy: Security and privacy of machine learning and AI applications

Abstract

Quantum machine learning deals with leveraging quantum theory with classic machine learning algorithms. Current research efforts study the advantages of using quantum mechanics or quantum information theory to accelerate learning time or convergence. Other efforts study data transformations in the quantum information space to evaluate robustness and performance boosts. This paper focuses on processing input data using randomized quantum circuits that act as quantum convolutions producing new representations that can be used in a convolutional network. Experimental results suggest that the performance is comparable to classic convolutional neural networks, and in some instances, using quantum convolutions can accelerate convergence.

Problem

• Developing quantum circuits for quantum convolutional autoencoders
• Comparing quantum convolution performance with classical neural networks
• Exploring quantum convolutions for accelerated learning convergence

Innovation

• Randomized quantum circuits for quantum convolutions
• Quantum convolutional autoencoder for data processing
• Quantum convolutions accelerate convergence performance