k-Nearest Neighbor Learning with Graph Neural Networks

k-nearest neighbor (kNN) is a widely used learning algorithm for supervised tasks. In practice, the main challenge when using kNN its high sensitivity to hyperparameter setting, including number of nearest neighbors k, distance function, and weighting function. To improve robustness hyperparameters, this study presents novel method based on graph neural network, named kNNGNN. Given training data, learns task-specific rule in an end-to-end fashion by means network that takes instance predict label instance. The functions are implicitly embedded within network. For query instance, prediction obtained performing search from data create passing it through effectiveness proposed demonstrated various benchmark datasets classification regression