QESK: Quantum-based Entropic Subtree Kernels for Graph Classification
This work addresses graph classification problems, offering a novel kernel that improves accuracy for tasks involving complex graph structures, though it appears incremental as it builds on quantum walks and Weisfeiler-Lehman algorithms.
The paper tackles graph classification by proposing the Quantum-based Entropic Subtree Kernel (QESK), which uses quantum walks and entropic subtree representations to capture structural characteristics and address limitations of existing kernels, resulting in significant outperformance over state-of-the-art methods in experiments.
In this paper, we propose a novel graph kernel, namely the Quantum-based Entropic Subtree Kernel (QESK), for Graph Classification. To this end, we commence by computing the Average Mixing Matrix (AMM) of the Continuous-time Quantum Walk (CTQW) evolved on each graph structure. Moreover, we show how this AMM matrix can be employed to compute a series of entropic subtree representations associated with the classical Weisfeiler-Lehman (WL) algorithm. For a pair of graphs, the QESK kernel is defined by computing the exponentiation of the negative Euclidean distance between their entropic subtree representations, theoretically resulting in a positive definite graph kernel. We show that the proposed QESK kernel not only encapsulates complicated intrinsic quantum-based structural characteristics of graph structures through the CTQW, but also theoretically addresses the shortcoming of ignoring the effects of unshared substructures arising in state-of-the-art R-convolution graph kernels. Moreover, unlike the classical R-convolution kernels, the proposed QESK can discriminate the distinctions of isomorphic subtrees in terms of the global graph structures, theoretically explaining the effectiveness. Experiments indicate that the proposed QESK kernel can significantly outperform state-of-the-art graph kernels and graph deep learning methods for graph classification problems.