Feature Selection and Classification Using Age Layered Population Structure Genetic Programming

The curse of dimensionality is a major problem in the fields of machine learning, data mining and knowledge discovery. Exhaustive search for the most optimal subset of relevant features from a high dimensional dataset is NP hard. Sub–optimal population based stochastic algorithms such as GP and GA are good choices for searching through large search spaces, and are usually more feasible than exhaustive and deterministic search algorithms. On the other hand, population based stochastic algorithms often suffer from premature convergence on mediocre sub–optimal solutions. The Age Layered Population Structure (ALPS) is a novel metaheuristic for overcoming the problem of premature convergence in evolutionary algorithms, and for improving search in the fitness landscape. The ALPS paradigm uses an age–measure to control breeding and competition between individuals in the population. This thesis uses a modification of the ALPS GP strategy called Feature Selection ALPS (FSALPS) for feature subset selection and classification of varied supervised learning tasks. FSALPS uses a novel frequency count system to rank features in the GP population based on evolved feature frequencies. The ranked features are translated into probabilities, which are used to control evolutionary processes such as terminal–symbol selection for the construction of GP trees/sub-trees. The FSALPS metaheuristic continuously refines the feature subset selection process whiles simultaneously evolving efficient classifiers through a non–converging evolutionary process that favors selection of features with high discrimination of class labels. We investigated and compared the performance of canonical GP, ALPS and FSALPS on high–dimensional benchmark classification datasets, including a hyperspectral image. Using Tukey’s HSD ANOVA test at a 95% confidence interval, ALPS and FSALPS dominated canonical GP in evolving smaller but efficient trees with less bloat expressions. FSALPS significantly outperformed canonical GP and ALPS and some reported feature selection strategies in related literature on dimensionality reduction.

Feature Selection and Classification Using Age Layered Population Structure Genetic Programming

The curse of dimensionality is a major problem in the fields of machine learning, data mining and knowledge discovery. Exhaustive search for the most optimal subset of relevant features from a high dimensional dataset is NP hard. Sub–optimal population based stochastic algorithms such as GP and GA are good choices for searching through large search spaces, and are usually more feasible than exhaustive and determinis- tic search algorithms. On the other hand, population based stochastic algorithms often suffer from premature convergence on mediocre sub–optimal solutions. The Age Layered Population Structure (ALPS) is a novel meta–heuristic for overcoming the problem of premature convergence in evolutionary algorithms, and for improving search in the fitness landscape. The ALPS paradigm uses an age–measure to control breeding and competition between individuals in the population. This thesis uses a modification of the ALPS GP strategy called Feature Selection ALPS (FSALPS) for feature subset selection and classification of varied supervised learning tasks. FSALPS uses a novel frequency count system to rank features in the GP population based on evolved feature frequencies. The ranked features are translated into probabilities, which are used to control evolutionary processes such as terminal–symbol selection for the construction of GP trees/sub-trees. The FSALPS meta–heuristic continuously refines the feature subset selection process whiles simultaneously evolving efficient classifiers through a non–converging evolutionary process that favors selection of features with high discrimination of class labels. We investigated and compared the performance of canonical GP, ALPS and FSALPS on high–dimensional benchmark classification datasets, including a hyperspectral image. Using Tukey’s HSD ANOVA test at a 95% confidence interval, ALPS and FSALPS dominated canonical GP in evolving smaller but efficient trees with less bloat expressions. FSALPS significantly outperformed canonical GP and ALPS and some reported feature selection strategies in related literature on dimensionality reduction.

GA approach for finding Rough Set decision rules based on bireducts

Feature selection plays an important role in knowledge discovery and data mining nowadays. In traditional rough set theory, feature selection using reduct - the minimal discerning set of attributes - is an important area. Nevertheless, the original definition of a reduct is restrictive, so in one of the previous research it was proposed to take into account not only the horizontal reduction of information by feature selection, but also a vertical reduction considering suitable subsets of the original set of objects. Following the work mentioned above, a new approach to generate bireducts using a multi--objective genetic algorithm was proposed. Although the genetic algorithms were used to calculate reduct in some previous works, we did not find any work where genetic algorithms were adopted to calculate bireducts. Compared to the works done before in this area, the proposed method has less randomness in generating bireducts. The genetic algorithm system estimated a quality of each bireduct by values of two objective functions as evolution progresses, so consequently a set of bireducts with optimized values of these objectives was obtained. Different fitness evaluation methods and genetic operators, such as crossover and mutation, were applied and the prediction accuracies were compared. Five datasets were used to test the proposed method and two datasets were used to perform a comparison study. Statistical analysis using the one-way ANOVA test was performed to determine the significant difference between the results. The experiment showed that the proposed method was able to reduce the number of bireducts necessary in order to receive a good prediction accuracy. Also, the influence of different genetic operators and fitness evaluation strategies on the prediction accuracy was analyzed. It was shown that the prediction accuracies of the proposed method are comparable with the best results in machine learning literature, and some of them outperformed it.

CEM - visualization and discovery in Email

This paper presents a lattice-based visual metaphor for knowledge discovery in electronic mail. It allows a user to navigate email using a visual lattice metaphor rather than a tree structure. By using such a conceptual multi-hierarchy, the content and shape of the lattice can be varied to accommodate any number of queries against the email collection. The system provides more flexibility in retrieving stored emails and can be generalised to any electronic documents. The paper presents the underlying mathematical structures, and a number of examples of the lattice and multi-hierarchy working with a prototypical email collection.

Efficient data mining based on formal concept analysis

Formal Concept Analysis is an unsupervised learning technique for conceptual clustering. We introduce the notion of iceberg concept lattices and show their use in Knowledge Discovery in Databases (KDD). Iceberg lattices are designed for analyzing very large databases. In particular they serve as a condensed representation of frequent patterns as known from association rule mining. In order to show the interplay between Formal Concept Analysis and association rule mining, we discuss the algorithm TITANIC. We show that iceberg concept lattices are a starting point for computing condensed sets of association rules without loss of information, and are a visualization method for the resulting rules.

Efficient mining of association rules based on formal concept analysis

Association rules are a popular knowledge discovery technique for warehouse basket analysis. They indicate which items of the warehouse are frequently bought together. The problem of association rule mining has first been stated in 1993. Five years later, several research groups discovered that this problem has a strong connection to Formal Concept Analysis (FCA). In this survey, we will first introduce some basic ideas of this connection along a specific algorithm, TITANIC, and show how FCA helps in reducing the number of resulting rules without loss of information, before giving a general overview over the history and state of the art of applying FCA for association rule mining.

Knowledge-based design cost estimation through extending industry foundation classes

In order to overcome divergence of estimation with the same data, the proposed digital costing process adopts an integrated design of information system to design the process knowledge and costing system together. By employing and extending a widely used international standard, industry foundation classes, the system can provide an integrated process which can harvest information and knowledge of current quantity surveying practice of costing method and data. Knowledge of quantification is encoded from literatures, motivation case and standards. It can reduce the time consumption of current manual practice. The further development will represent the pricing process in a Bayesian Network based knowledge representation approach. The hybrid types of knowledge representation can produce a reliable estimation for construction project. In a practical term, the knowledge management of quantity surveying can improve the system of construction estimation. The theoretical significance of this study lies in the fact that its content and conclusion make it possible to develop an automatic estimation system based on hybrid knowledge representation approach.

Developing a knowledge management system using an ontological approach in global organisation

This paper introduces an ontology-based knowledge model for knowledge management. This model can facilitate knowledge discovery that provides users with insight for decision making. The users requiring the insight normally play different roles with different requirements in an organisation. To meet the requirements, insights are created by purposely aggregated transnational data. This involves a semantic data integration process. In this paper, we present a knowledge management system which is capable of representing knowledge requirements in a domain context and enabling the semantic data integration through ontology modeling. The knowledge domain context of United Bible Societies is used to illustrate the features of the knowledge management capabilities.

An agent-based hybrid framework for database mining

While knowledge discovery in databases (KDD) is defined as an iterative sequence of the following steps: data pre-processing, data mining, and post data mining, a significant amount of research in data mining has been done, resulting in a variety of algorithms and techniques for each step. However, a single data-mining technique has not been proven appropriate for every domain and data set. Instead, several techniques may need to be integrated into hybrid systems and used cooperatively during a particular data-mining operation. That is, hybrid solutions are crucial for the success of data mining. This paper presents a hybrid framework for identifying patterns from databases or multi-databases. The framework integrates these techniques for mining tasks from an agent point of view. Based on the experiments conducted, putting different KDD techniques together into the agent-based architecture enables them to be used cooperatively when needed. The proposed framework provides a highly flexible and robust data-mining platform and the resulting systems demonstrate emergent behaviors although it does not improve the performance of individual KDD techniques.

Linear causal model discovery using the MML criterion

Determining the causal structure of a domain is a key task in the area of Data Mining and Knowledge Discovery.The algorithm proposed by Wallace et al. [15] has demonstrated its strong ability in discovering Linear Causal Models from given data sets. However, some experiments showed that this algorithm experienced difficulty in discovering linear relations with small deviation, and it occasionally gives a negative message length, which should not be allowed. In this paper, a more efficient and precise MML encoding scheme is proposed to describe the model structure and the nodes in a Linear Causal Model. The estimation of different parameters is also derived. Empirical results show that the new algorithm outperformed the previous MML-based algorithm in terms of both speed and precision.

Building agent-based hybrid intelligent systems

Many complex problems including financial investment planning, foreign exchange trading, knowledge discovery from large/multiple databases require hybrid intelligent systems that integrate many intelligent techniques including expert systems, fuzzy logic, neural networks, and genetic algorithms. However, hybrid intelligent systems are difficult to develop because they have a large number of parts or components that have many interactions. On the other hand, agents offer a new and often more appropriate route to the development of complex systems, especially in open and dynamic environments. In this paper, it is argued that agent technology is well snited for constructing hybrid intelligent systems (especially loosely coupled hybrid intelligent systems) through a successful case study. A great number of heterogeneous computing techniques/packages are easily integlated into the experimental system under a unifying agent framework, which implies that agent technology can greatly facilitate the construction of hybrid intelligent systems.

Optimization of algorithm to identification of duplicate tuples through similarity phonetic based on multithreading

Aiming to ensure greater reliability and consistency of data stored in the database, the data cleaning stage is set early in the process of Knowledge Discovery in Databases (KDD) and is responsible for eliminating problems and adjust the data for the later stages, especially for the stage of data mining. Such problems occur in the instance level and schema, namely, missing values, null values, duplicate tuples, values outside the domain, among others. Several algorithms were developed to perform the cleaning step in databases, some of them were developed specifically to work with the phonetics of words, since a word can be written in different ways. Within this perspective, this work presents as original contribution an optimization of algorithm for the detection of duplicate tuples in databases through phonetic based on multithreading without the need for trained data, as well as an independent environment of language to be supported for this. © 2011 IEEE.

A distributed, semiotic-inductive and human-Oriented approach to web-scale knowledge retrieval

Web-scale knowledge retrieval can be enabled by distributed information retrieval, clustering Web clients to a large-scale computing infrastructure for knowledge discovery from Web documents. Based on this infrastructure, we propose to apply semiotic (i.e., sub-syntactical) and inductive (i.e., probabilistic) methods for inferring concept associations in human knowledge. These associations can be combined to form a fuzzy (i.e.,gradual) semantic net representing a map of the knowledge in the Web. Thus, we propose to provide interactive visualizations of these cognitive concept maps to end users, who can browse and search the Web in a human-oriented, visual, and associative interface.

Algorithm for the detection of outliers based on the theory of rough sets

Outliers are objects that show abnormal behavior with respect to their context or that have unexpected values in some of their parameters. In decision-making processes, information quality is of the utmost importance. In specific applications, an outlying data element may represent an important deviation in a production process or a damaged sensor. Therefore, the ability to detect these elements could make the difference between making a correct and an incorrect decision. This task is complicated by the large sizes of typical databases. Due to their importance in search processes in large volumes of data, researchers pay special attention to the development of efficient outlier detection techniques. This article presents a computationally efficient algorithm for the detection of outliers in large volumes of information. This proposal is based on an extension of the mathematical framework upon which the basic theory of detection of outliers, founded on Rough Set Theory, has been constructed. From this starting point, current problems are analyzed; a detection method is proposed, along with a computational algorithm that allows the performance of outlier detection tasks with an almost-linear complexity. To illustrate its viability, the results of the application of the outlier-detection algorithm to the concrete example of a large database are presented.

Finding negative event oriented patterns in long temporal sequences

Pattern discovery in a long temporal event sequence is of great importance in many application domains. Most of the previous work focuses on identifying positive associations among time stamped event types. In this paper, we introduce the problem of defining and discovering negative associations that, as positive rules, may also serve as a source of knowledge discovery. In general, an event-oriented pattern is a pattern that associates with a selected type of event, called a target event. As a counter-part of previous research, we identify patterns that have a negative relationship with the target events. A set of criteria is defined to evaluate the interestingness of patterns associated with such negative relationships. In the process of counting the frequency of a pattern, we propose a new approach, called unique minimal occurrence, which guarantees that the Apriori property holds for all patterns in a long sequence. Based on the interestingness measures, algorithms are proposed to discover potentially interesting patterns for this negative rule problem. Finally, the experiment is made for a real application.