Quantum Diffusion Convolution Kernels on Graphs

Explore the innovative Quantum Diffusion Convolution (QDC) operator for graph convolutional neural networks in this 59-minute lecture by Thomas Markovich at the Alan Turing Institute. Delve into the concept of generalized diffusion of input features on graphs and how QDC effectively rewires the graph based on vertex occupation correlations. Learn about the multiscale variant that combines messages from the QDC operator and the traditional combinatorial Laplacian. Examine the spectral dependence of homophily and the significance of quantum dynamics in creating a bandpass filter. Discover how QDC improves predictive performance on widely used benchmark datasets compared to similar methods, and gain insights into the potential applications of this novel approach in graph-based machine learning tasks.

Thomas Markovich - Quantum Diffusion Convolution Kernels on Graphs