“…The second data set is a mixed database consisting of 10,000 images, and we use it for the k-NN search experiment. More precisely, it is composed of 1,000 airplane images, 3,000 baseball images, 1,200 flower images, 3,000 bird images, 800 motorcycle images, and 1,000 guitar images, which are collected from the Web [6]. In k-NN search, the parameter p of WSLPM is not used, and we thus do not specify its value in the experimental result.…”