What Is the Nearest Neighbor in High Dimensional Spaces

摘要: Nearest neighbor search in high dimensional spaces is an interesting and important problem which relevant for a wide variety of novel database applications. As recent results show, however, the very di cult one, not only with regards to performance issue but also quality issue. In this paper, we discuss identify new generalized notion nearest as space. contrast previous approaches, our does treat all dimensions equally uses criterion select (projections) respect given query. example useful criterion, rate how well data clustered around query point within selected projection. We then propose e cient ective algorithm solve problem. Our experiments based on number real synthetic sets show that approach provides insights into nature