Tianqing Zhang

Mining Predicate Association Rule by Gene Expression Programming

Gene expression programming (GEP) is a new technique in genetic computing introduced in 2001. Association rule mining is a typical task in data mining. In this article, a new concept called Predicate Association (PA) is introduced and a new method to discover PA by GEP, called PAGEP (mining Predicate Association by GEP), is proposed. Main results are: (1) The inherent weaknesses of traditional association (TA) are explored. It is proved that TA is a special case of PA. (2) The algorithms for mining PAR, decoding chromosome and fitness are proposed and implemented. (3) It is also proved that gene decoding procedure always success for any well-defined gene. (4) Extensive experiments are given to demonstrate that PAGEP can discover some association rule that cannot be expressed and discovered by traditional method.

Distance guided classification with gene expression programming

Gene Expression Programming (GEP) aims at discovering essential rules hidden in observed data and expressing them mathematically. GEP has been proved to be a powerful tool for constructing efficient classifiers. Traditional GEP-classifiers ignore the distribution of samples, and hence decrease the efficiency and accuracy. The contributions of this paper include: (1) proposing two strategies of generating classification threshold dynamically, (2) designing a new approach called Distance Guided Evolution Algorithm (DGEA) to improve the efficiency of GEP, and (3) demonstrating the effectiveness of generating classification threshold dynamically and DGEA by extensive experiments. The results show that the new methods decrease the number of evolutional generations by 83% to 90%, and increase the accuracy by 20% compared with the traditional approach.

Web Document Filtering Technique Based on Natural Language Understanding

Web document filtering is an important aspect in information security. The traditional strategy based on simple keywords matching often leads to low accuracy of discrimination. This article proposes the method FNLU (Filtering based on Natural Language Understanding) including the algorithms for Extracting Typical Phrase, Calculating feature vector, Mining threshold vector, Objective judging and Subjective judging. The experimental result shows that the algorithms are efficient.

Mining Class Contrast Functions by Gene Expression Programming

Finding functions whose accuracies change significantly between two classes is an interesting work. In this paper, this kind of functions is defined as class contrast functions. As Gene Expression Programming (GEP) can discover essential relations from data and express them mathematically, it is desirable to apply GEP to mining such class contrast functions from data. The main contributions of this paper include: (1) proposing a new data mining task --- class contrast function mining, (2) designing a GEP based method to find class contrast functions, (3) presenting several strategies for finding multiple class contrast functions in data, (4) giving an extensive performance study on both synthetic and real world datasets. The experimental results show that the proposed methods are effective. Several class contrast functions are discovered from the real world datasets. Some potential works on class contrast function mining are discussed based on the experimental results.

O(1) time algorithm on BSR for constructing a random binary search tree

Constructing a random binary search tree with n nodes needs /spl theta/(nlog n) time on RAM, and /spl omega/(log n) time on n-processor EREW, CREW, or CRCW PRAM. We propose an O(1) time algorithm on n-processor BSR PRAM for the problem, which is the first constant time solution to the problem on any model of computation.

Discovering Tendency Association between Objects with Relaxed Periodicity and its Application in Seismology

Relaxed periodicity is proposed to describe loose-‘cyclic behavior of objects while allowing uneven stretch or shrink on time axis, limited noises, and inflation/deflation of attribute values. The techniques to mine the relaxed periodicity and the association between objects with relaxed periodicity are studied. The proposed algorithms are tested by the data in the Seismic database of Annin River area, and its results are interesting to seismology.

Mining contrast inequalities in numeric dataset

Finding relational expressions which exist frequently in one class of data while not in the other class of data is an interesting work. In this paper, a relational expression of this kind is defined as a contrast inequality. Gene Expression Programming (GEP) is powerful to discover relations from data and express them in mathematical level. Hence, it is desirable to apply GEP to such mining task. The main contributions of this paper include: (1) introducing the concept of contrast inequality mining, (2) designing a two-genome chromosome structure to guarantee that each individual in GEP is a valid inequality, (3) proposing a new genetic mutation to improve the efficiency of evolving contrast inequalities, (4) presenting a GEP-based method to discover contrast inequalities, (5) giving an extensive performance study on real-world datasets. The experimental results show that the proposed methods are effective. Contrast inequalities with high discriminative power are discovered from the real-world datasets. Some potential works on contrast inequality mining are discussed.

An Effective Microarray Data Classifier Based on Gene Expression Programming

Applying data mining algorithms to microarray data analysis is an interesting and promising work. Gene Expression Programming (GEP) is a new development of evolution computation. GEP performs global search and discover the classification discriminant with high accuracy. However, it is undesirable to apply GEP on microarray classification directly, since the evolution efficiency of GEP is low when the number of attributes of training data is huge. To solve this problem, the main contributions of this paper include: (1) analyzing the difficulties of applying GEP to classifying microarray data directly, (2) designing a novel method to select GEP terminals from genes of microarray data, (3) proposing a method of constructing GEP classifier committee to improve the classification accuracy, (4) demonstrating the effectiveness of proposed algorithms by extensive experiments on several microarray data. Compared with some typical classification methods, the accuracy is increased as high as 10.46% in average.

Construct Distributed RFID Data Warehouse Based On Concept Hierarchy

Radio frequency identification (RFID) applications play an important role in business activities nowadays. With the rapidly increasing volume of RFID data, the traditional centralized data warehouses (CDW) can not process RFID efficiently. To solve the problem, this paper proposes a novel model of RFID distributed data warehouse based on concept hierarchy (RFID-CFIDDW). The main contributions include: 1) proposing a new warehousing model, 2) introducing the implementation of the new model, 3) demonstrating the utility and feasibility of the new model via experiments.

Meta Galaxy: A Flexible and Efficient Cube Model for Data Retrieval in OLAP

OLAP is widely used in data analysis. The existing design models, such as star schema and snowflake schema, are not flexible when the data model is changed. For example, the task for inserting a dimension may involve complex operations over model and application implementation. To deal with this problem, a new cube model, called Meta Galaxy, is proposed. The main contributions of this work include: (1) analyzing the shortcoming of traditional design method, (2) proposing a new cube model which is flexible for dimension changes, and (3) designing an index structure and an algorithm to accelerate the cube query. The time complexity of query algorithm is linear. The extensive experiments on the real application and synthetic dataset show that Meta Galaxy is effective and efficient for cube query. Specifically, our method decreases the storage size by 95.12%, decreases the query time by 89.89% in average compared with SQL Server 2005, and has good scalability on data size.