M. Khan, Q. Ding, and W. Perrizo, "K-nearest Neighbor Classification of Spatial Data Streams Using P-trees," Pacific-Asian Conference on Knowledge Discovery and Data Mining, PAKDD-2002.  Home/Search   Document Details and Download   Summary   Related Articles   Check

No context found.

M Khan, Q Ding, W Perrizo, `K-nearest Neighbor Classification on Spatial Data Stream Using PTrees ', PAKDD-02, Taipei, pp.517-528, May 2002.

No context found.

Maleq Khan, Qin Ding, William Perrizo, "K-Nearest Neighbor classification of spatial data streams using P-trees", PAKDD-2002.

No context found.

M. Khan, Q. Ding, and W. Perrizo, "K-nearest Neighbor Classification of Spatial Data Streams Using Ptrees, " PAKDD-2002.

No context found.

M. Khan, Q. Ding, W. Perrizo, "K-Nearest Neighbor classification of spatial data streams using P-Trees", PAKDD-2002.

....bSQ file. Fast P tree operations, especially the fast AND operation, provide possibilities for efficient data mining. We have proposed P tree based efficient data mining algorithms, such as association rule mining [DDP02b] decision tree induction [DDP02a] and k nearest neighbor classification [KDP02]. In this dissertation, we will focus on P tree based association rule mining of RSI data. The basic idea of our approach is that, by using P trees during the mining process, database scans can be replaced by appropriate P tree operations, which are much faster. In addition, P trees facilitate ....

....environmental analysis and control. Besides association rule mining, we can also perform other data mining applications, such as classification and clustering. We proposed a decision tree induction algorithm based on P trees [DDP02a] and a k nearest neighbor classification algorithm using P trees [KDP02]. We believe that there are open areas in data mining which will benefit from using P trees, such as sequential pattern mining and data mining on data streams. 121 ....

M. Khan, Q. Ding, and W. Perrizo, "K-nearest Neighbor Classification on Spatial Data Stream Using P-trees," Proceedings of the PAKDD, Taipei, Taiwan, May 2002, pp. 517-528. 128

....00 represents a pure0 quadrant and 01 represents a mixed quadrant. To simplify, we use 1 for pure1, 0 for pure0, and m for mixed. This is illustrated in the rd part of Figure 1. P tree algebra contains operators, AND, OR, NOT and XOR, which are the pixel by pixel logical operations on P trees [3]. The NOT operation is a straightforward translation of each count to its quadrantcomplement. The AND and OR operations are showninFigure2. Figure2.P treeAlgebra The basic P trees can be combined using simple logical operations to produce P trees for the original values (at any level of ....

....n i q i i i q y x w Y X d ) If q = 2, this gives the Euclidean function. If q = 1, it gives the Manhattan distance, which is 1 1 ) i i i y x Y X d . Ifq= it gives the max function i n y x Y X d = max ) We proposed a metric using P trees, called HOBBit [3]. The HOBBit metric measures distance based on the most significant consecutive bit positions starting from the left (the highest order bit) The HOBBit similarity between two integers A and B is defined by S H (A, B) max s 0 i s # a i =b i . eq. 1 ) where a i and b i are the i bits ....

Maleq Khan, Qin Ding, William Perrizo, "kNearest Neighbor Classification on Spatial Data Streams Using P-Trees", PAKDD 2002.

....set or arbitrarily ignore some of them. Other methods find the exact k nearest neighbor set, since that is easiest using traditional techniques although it is clear that allowing some samples at that distance to vote and not others, will skew the result. Instead, the P tree based KNN approach [4] builds a closed nearest neighbor set (closed KNN) that is, we include all of the boundary neighbors. The inductive definition of the closed KNN set is given below. a) If x KNN, then x closed KNN (b) If x closed KNN and d(T,y) d(T,x) then y closedKNN, where, d(T,x) is the distance ....

....of the closed KNN set is given below. a) If x KNN, then x closed KNN (b) If x closed KNN and d(T,y) d(T,x) then y closedKNN, where, d(T,x) is the distance of x from target T. c) Closed KNN does not contain any tuple which cannot be produced by steps a and b. Experimental results [4] show closed KNN yields higher classification accuracy than KNN does. If there are many tuples on the boundary, inclusion of some but not all of them skews the voting mechanism. The P tree implementation requires no extra computation to find the closed KNN. Our neighborhood expansion mechanism ....

[Article contains additional citation context not shown here]

Maleq Khan, Qin Ding, William Perrizo, "k-Nearest Neighbor Classification on Spatial Data Streams Using PTrees ", PAKDD 2002.

....algorithm that uses k medoids related ideas without incurring the high time complexity. In Birch, data are broken up into local clustering features and then combined into CF Trees. In contrast to Birch we determine cluster centers that are defined globally. We represent data in the form of P trees [10, 11, 12, 13, 14], which are efficient both in the storage requirements and the time complexity of computing global counts. The rest of the paper is organized as follows. In section 2 we analyze the reasons for high time complexity of k medoids based algorithms and show how this complexity can be avoided using a ....

....requires us to have a fast method of finding data points based on their feature attribute values. Density based algorithms such as DENCLUE achieve this goal by saving data in a special data structure that allows referring to neighbors. We use a data structure, namely a Peano Count Tree (or P tree) [10, 11, 12, 13, 14] that allows fast calculation of counts of data points based on their attribute values. 4.1 A Summary of P Tree Features Many types of data show continuity in dimensions that are not themselves used as data mining attributes. Spatial data that is mined independently of location will consist of ....

[Article contains additional citation context not shown here]

Maleq Khan, Qin Ding, William Perrizo, "K-Nearest Neighbor Classification of Spatial Data Streams using Ptrees ", PAKDD-2002.

No context found.

M. Khan, Q. Ding, and W. Perrizo, "K-nearest Neighbor Classification of Spatial Data Streams Using P-trees," Pacific-Asian Conference on Knowledge Discovery and Data Mining, PAKDD-2002.

No context found.

M. Khan, Q. Ding, and W. Perrizo, "K-nearest Neighbor Classification of Spatial Data Streams Using P-trees," PAKDD-2002.

No context found.

M. Khan, Q. Ding, and W. Perrizo, "K-Nearest Neighbor Classification of Spatial Data Streams using P-trees," Pacific Asian Conference on Knowledge Discovery and Data Mining (PAKDD-2002.

No context found.

M. Khan, Q. Ding, and W. Perrizo, "K-nearest Neighbor Classification of Spatial Data Streams Using P-trees," Pacific Asian Conf. on Knowledge Discovery and Data Mining (PAKDD-2002.

No context found.

M. Khan, Q. Ding, and W. Perrizo, "K-nearest Neighbor Classification of Spatial Data Streams Using P-trees," Pacific Asian Conf. on Knowledge Discovery and Data Mining (PAKDD-2002.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC