Citation Context

... It is amongst the simplest of all machine learning algorithms [5]. BM25 is a widely used similarity measurement in KNN, through which the k nearest training samples for a given test sample are found =-=[17, 18]-=-(see Sec. 3.1 for details). As the KNN classifier requires storing and searching the whole training set which becomes too costly when this set grows large, many researchers have attempted to reduce th...