Publications by authors named "Alain Pirotte"
Article Synopsis
- The paper conducts a thorough survey and empirical evaluation of seven graph kernels and two related similarity matrices, collectively referred to as "kernels on graphs."
- It introduces various kernels like the exponential diffusion kernel and the regularized commute-time kernel, examining their effectiveness in collaborative-recommendation tasks and semi-supervised classification on multiple datasets.
- Results indicate that the regularized commute-time and Markov diffusion kernels excel in performance, with the regularized Laplacian kernel also showing strong results in the evaluated applications.
