Bayesian Quantum Orthogonal Neural Networks for Anomaly Detection

Learn about a novel approach combining Bayesian learning with quantum-inspired machine learning techniques for industrial anomaly detection in this 20-minute conference talk. Explore how researchers developed orthogonal quantum versions of 3D convolutional neural networks that successfully identify defects in 3D objects, with Bayesian methods providing uncertainty quantification and orthogonal weight matrices enabling smoother training processes. Discover the practical implementation of these models through hardware experiments conducted on IBM's 127-qubit Brisbane quantum device, examining the effects of quantum noise and measurement limitations on performance. Gain insights into the intersection of quantum computing, machine learning, and industrial quality control applications, including the challenges and opportunities of deploying quantum-enhanced anomaly detection systems in real-world manufacturing environments.