Three related types of multi-value association patterns

Mining patterns involving multiple values that are significantly relevant is a difficult but very important problem that crosses many disciplines. Multi-value association patterns, which generalize sequential pattern, are sets of associated values extracted from sampling outcomes of a random N-tuple. Because they are value patterns from multiple variables, they are more descriptive than their corresponding variable patterns. Hence, they are also easier to interpret. Normally, they can be detected by statistical testing if the occurrence of a pattern event is significantly deviated from the expected according to a prior model or null hypothesis. In this paper, we consider three related types of multi-value association patterns including high-order pattern (HOP), consigned pattern (CP), and nested high-order pattern (NHOP). We further evaluate the nested high-order pattern and its relationships to the others using experiments.

• Type: Conference
• Year: 2008
• Where: ICPR
• Authors: Thomas W. H. Lui, David K. Y. Chiu
• Keywords: Computer Vision, Corresponding Variable Patterns, ICPR 2008, Multi-value Association Patterns, Sequential Pattern