In classification problems, the boundary between the regions of different classes
Hamming distance
for two feature vectors with discrete values, the number of features for which two vectors differ
Euclidean distance
for two feature vectors with real values, the sum of squared differences between the features in the two vectors
Manhattan distance
for two feature vectors with real values, the sum of absolute differences between the features in the two vectors
Tuning set
a set of data points, separated from training or test data. It is used to evaluate the performance of machine learning models trained on the training set with certain input configurations (e.g., certain number of neighbors for k-nearest neighbors). Then the configuration with the best performance on the tuning set is selected. Finally, a model is trained on the training set with the selected configuration, and then evaluated on the test set.