Fig. 1): The rule creation algorithm creates an initial fuzzy partitioning for each variable. This is given by a xed number of equally distributed tri

Size: px
Start display at page:

Download "Fig. 1): The rule creation algorithm creates an initial fuzzy partitioning for each variable. This is given by a xed number of equally distributed tri"

Transcription

1 Some Approaches to Improve the Interpretability of Neuro-Fuzzy Classiers Aljoscha Klose, Andreas Nurnberger, and Detlef Nauck Faculty of Computer Science (FIN-IWS), University of Magdeburg Universitatsplatz 2, D Magdeburg, Germany Tel , Fax: URL: Abstract Neuro-fuzzy classication systems make it possible to obtain a suitable fuzzy classier by learning from data. Nevertheless, in some cases the derived rule base is hard to interpret. In this paper we discuss some approaches to improve the interpretability of neuro-fuzzy classication systems. We present modied learning strategies to derive fuzzy classication rules from data, and some methods to simplify the found rule base to improve the interpretability of the resulting fuzzy system. 1 Introduction Fuzzy rules are a well-suited representation of classication knowledge. Their application is rather intuitive and understandable. The abstraction from gures to linguistic variables eases the readability and interpretability of the rule base. Automatic induction of fuzzy rules from data is therefore an interesting topic. One approach to automatic rule generation is based on neuro-fuzzy systems, that use learning algorithms derived from neural network theory to generate fuzzy rules [1, 5, 12]. Unfortunately, the result of automatic rule induction is not always as easy to interpret as arulebasebuiltby human experts. In this paper we discuss some of the problems and approaches to improve the interpretability of neuro-fuzzy classication systems. We present modied learning strategies to derive fuzzy classication rules from data, and some methods to simplify the found rule base to improve the interpretability for the user. For the analyses of the presented approaches we used an implementation of the NEFCLASS (NEuro Fuzzy CLASSication) model [6], a neuro-fuzzy model for data analysis, since it was designed as an interactive classication tool. One of our goals is to develop algorithms that can learn automatically, but also allow the user to inuence the learning and classication process, e.g. by initializing the system, and modifying or extracting knowledge. The next section describes the NEFCLASS system and its two-phase mechanism of rule creation and ne tuning. Sect. 2 shows problems of automatic rule generation and describes our approaches to tackle them. Experimental evaluation and conclusions are given in Sect. 3 and The NEFCLASS Model NEFCLASS can be viewed as a special 3-layer feed-forward neural network. The units in this network use t-norms or t-conorms instead of the activation functions common in neural networks. The rst layer represents input variables, the middle (hidden) layer represents fuzzy rules and the third layer represents output variables. Fuzzy sets are encoded as (fuzzy) connection weights. Thus NEFCLASS can always (i.e. before, during and after learning) be interpreted as a system of fuzzy classication rules like where the 1 ::: n are fuzzy sets. if x1 is 1 and x2 is 2 and ::: and xn is n then pattern (x1 x2 ::: xn) belongs to class i, It is possible to create a NEFCLASS system from scratch using training data, or to initialize it by prior knowledge in form of fuzzy rules. If a classier is learned from data, two phases are needed (see left side of Acknowledgment: The research presented in this paper is partly funded by DFG contract KR 521/3-1.

2 Fig. 1): The rule creation algorithm creates an initial fuzzy partitioning for each variable. This is given by a xed number of equally distributed triangular membership functions. The combination of the fuzzy sets forms a grid in the data space, i.e. equally distributed overlapping hyper boxes. Then the training data is processed, and those clusters, that cover areas where data is located are added as rules into the rule base of the classier. After the rule base has been created, the membership functions are tuned by a simple heuristic. For each rule a classication error is determined and used to modify that membership function that is responsible for the rule activation (i.e. which yields the minimal membership degree of all fuzzy sets in the rule's antecedent). The modication results in shifting the fuzzy set, and enlarging or reducing its support, such that a larger or smaller membership degree is obtained depending on the current error. The learning procedure takes into account the semantic properties of the underlying fuzzy system. This results in constraints on possible modications applicable to the system parameters. NEFCLASS allows to impose dierent restrictions on the learning algorithm, e.g. membership functions must not pass its neighbors, must stay symmetrical, or intersect at 0.5 [7]. 2 Improving Neuro-Fuzzy Classication One major problem of NEFCLASS and other fuzzy rule inducing systems is, that they tend to nd large numbers of rules and therefore interpretability (aswell as generalizing ability and learning stability) cannot be guaranteed. This is mainly because the rule creation process does not look for larger structures in the data, but for a local covering of the input space with hyperboxes. For datasets of higher dimensionality there are huge numbers of possible rules. The Wisconsin breast cancer dataset of the UCI machine learning repository [4] has 9 input variables. Using all of them and choosing a partitioning of the input ranges into 5 linguistic variables each, results in 5 9 = possible rules. Thus, most of the about 700 cases in the dataset generate rules which cover only the generating case. This will obviously deliver a poor classication on unseen data. If we dene interpretability as the possibility togaininsight in the structures of the data from looking at the rules, this interpretability willbevery low. Finding structure in the data means abstraction from single cases by describing it in more general rules. Thus it is a common target for generality andinterpretability to reduce the number of rules. In [8] some (semi-automatic) pruning methods were presented, which can be used interactively during the learning process. One basic principle of these methods is to prune the rules by deleting hyperboxes that { by some measure { do not essentially contribute to the classication result. As the rules are fuzzy and thus hyperboxes overlap, the neighboring rules can (hopefully) classify cases formerly classied by now deleted rules. Successive runs of the second learning phase try to enlarge the remaining fuzzy rules (hyperboxes), such that the input space is completely covered again [8]. Rule simplication is done by iteratively applying learning phase II and pruning (see Fig. 1). Problems arise when the data sparsely covers the input space or the initial hyperboxes are small. In the breast cancer example mentioned above many rules classify only one data point. Therefore, the optimizing learning phase II has problems to adapt the fuzzy sets correctly due to the lack of statistical information. The pruning methods make rather crude changes to the rule base by rule deleting rules and thus strongly change the classication. We therefore show an alternative approach that reduces the number of rules, but tries to keep the changes in classication performance small. The basic principles are the merging of neighboring hyperboxes (i.e. rules), deletion of redundant fuzzy sets and choosing an appropriate rule base format. Merging of Rules If hyperboxes are adjacent in input space and have identical consequents, they can be joined with no or little loss of information. Thus, initial hyperboxes are enlarged and joined rules are combined to a single one. Joining rules that are adjacent in one dimension can prohibit joins in other dimensions. Therefore the task is to nd the best merge, i.e. merging the most rules with a minimal loss of classication performance. Unfortunately, this search is NP-complete, so we have to make use of some heuristics. A usual strategy is greedy searching, i.e. promising joins combining many rules are made rst. If, for example, a rule can be joined with an adjacent rule in one dimension, but with two or more rules along another dimension, the latter is preferred. A special case are rules, that are neighboring along one input dimension, predict the same class, and cover the complete input range. This input feature can assume arbitrary values, and can therefore be ignored in the joined rule. In the fuzzy rule base this is done by setting this input to the linguistic term

3 Figure 1: Structure of the learning algorithm in a modied NEFCLASS system don't care, which is associated with a constant membership function equal to 1. Another even more promising situation can be seen as a special case of the above-mentioned scenario: If for all rules the same input can be ignored, it can completely be removed from the input dataset. This is desirable, as it reduces very eectively the number of rules and the input complexity of the dataset. According to our greedy approach, we merge rules in 3 phases: First, we search explicitly for inputs that can be ignored in all rules, then for inputs, that can be ignored in some rules, and then for all remaining merging possibilities. We call these 3 phases input pruning on dataset level, input pruning on rule level and (simple) rule merging. Usually, the rst two phases will be applied after initial rule base creation, but they may also be used in the pruning loop (see Fig. 1). We willnow describe the phases in more detail, beginning with pruning on dataset level. Input Pruning on Data Set Level Inspired by the top-down induction of decision trees [9, 10], we start with all rules merged and split them iteratively. That means we start with no inputs used for classication. Then iteratively the best input according to an appropriate selection measure is added to the set of used inputs. Previously merged rules that dier in the newly selected variable are split accordingly. This procedure is repeated until the expected classication error reaches a minimum or falls beneath a threshold given by the user. All inputs not belonging to the set of used inputs are set to don't care and all superuous rules are removed. This greedy approach strongly depends on the attribute selection measure. We decide to use a measure based on the minimum description length principle (MDL, [11]), which in the past turned out to be well suited for the induction of decision trees [2]. MDL is equivalent to maximum likelihood for the estimation of models with a xed number of free parameters, but additionally oers a possibility to compare objectively between models of dierent complexity. Intuitively, in a classication case MDL yields good values (short description lengths) for models with few parameters, that separate the classes well. On the other hand it punishes models with many parameters. Thus it helps to prevent over-tting. The basis for the application of MDL is a crisp statistic built from the dataset. It counts how many cases of each class each rule classies, i.e. each case is counted for the rule which has the highest activation. From this statistic it can be calculated, which rules and which cases fall together, if a set of inputs is ignored. From this the expected error can be calculated and the separation can be measured by MDL. Due to lack of space we refer the reader to [2] for a formal discussion. Input Pruning on Rule Level The next step is pruning on rule level, i.e. the search for a set of rules that are adjacent in one dimension, independently from other rules. If they cover a certain input range and do not make contradicting predictions, they are merged by ignoring the corresponding input. In contrast to the last section, this merging works bottom-up. Again a greedy search guided by MDL is applied. This time, no error limit is set. Instead of this, the message length for the unmerged and merged rules are calculated according to [2]. The merging is only done, if this decreases the MDL measure.

4 Simple Rule Merging The third step is simple rule merging. This can be done on several levels: Rules with neighboring fuzzy sets can be merged by using a trapezoidal fuzzy set in the antecedent: The interpretability is not aected, since usually a new linguistic term can be given which combines both fuzzy sets. The membership values of data points between the merged fuzzy sets (and thus classication) will be aected, as the trapezoidal fuzzy set has higher membership values for the inner region. However, if the rules are merged because they describe a common region for a class in the data set, this even is desirable. Rules can be merged on a logical level: The rule base is given in disjunctive normal form. More compact representations can be given using Boolean transformations. This is only applicable in the post-processing phase before the nal rule base is given to the user, since the rules' ring strengths could be modied in an undesired way. Nearly identical fuzzy sets can be pruned: During optimization fuzzy sets often become very similar. They can be replaced by a single fuzzy set if they are equal according to a similarity measure (see, for example, [3]). Sorting and Grouping Rule Output Our approaches can be divided into pruning in the learning loop and post-processing of the rule base. The pruning loop aims at increasing the interpretability as well as supporting learning and enable more powerful representations. Post-processing only aims at providing a more readable rule base. Boolean transformations on a logical level is an example of such a kind of post-processing. Additional help can be given to the user by choosing an appropriate output format. Rules that cannot be merged, but nevertheless describe adjacent regions, should be grouped in output. In this way the user is guided in interpreting the classication. 3 Experimental Evaluation As an example for our extensions to NEFCLASS we use the \Wisconsin Breast Cancer" data set [13]. This data set contains 699 cases distributed into two classes (benign and malign). We excluded 16 cases with missing values, and split the dataset into training and test data. Initial rule creation using 5 fuzzy sets per variable creates 166 rules. To demonstrate the pruning strategies the initial number of rules was not restricted. This means that \best per class" rule learning [7] to reduce IF Input 2 IS very small AND Input 6 IS (very small, small OR medium) THEN Class IS Benign IF Input 2 IS small AND Input 6 IS very small THEN Class IS Benign IF Input 6 IS (large OR very large) IF Input 6 IS (small or medium) AND Input 2 IS (small, medium, large OR very large) IF Input 6 IS very small AND Input 2 IS (medium, large OR very large) Table 1: Resulting rules for breast cancer data set

5 the number of rules right away was not used in this case. After training the learned rule base causes only one classication error. However, on the test set it performs badly with 132 errors (38%). 127 of these errors were ambiguous classications, i.e. cases, where no rule res. Obviously the hyperboxes are much too small and scattered. Therefore, we next applied the merging/pruning strategies presented in this paper to the rulebase. The rst pruning phase with 5% misclassications as stopping criterion uses only two inputs to obtain 9 (2.6%) errors on the training set and 18 errors on the test set. There are only 25 rules left. The next pruning phase reduces this number by 8 to a remainder of 17 rules. The classication performance is not aected by this merging. Tab. 1 shows the ordered and logically merged rule output. The last step has not yet been implemented and has therefore been done manually. It should be noted that the number of rules can be further decreased by the other pruning techniques implemented in NEFCLASS [8]. 4 Conclusions The presented approaches can be used as automatic as well as interactive pruning methods. Some of them are not limited to neuro-fuzzy classication. They may be applied to all kinds of classiers, and some (Sect. 2) even do not need a representation by rules. Nonetheless, in combination with adapted learning methods, improvementofinterpretability and learning stabilitycanbeachieved. Together with other pruning methods recently integrated in NEFCLASS [8] the automatic part (the machine's support) in the interactive process of data analysis can be increased. The original NEFCLASS software for UNIX (written in C++, with a user interface in TCL/TK) that was used to obtain the results described in this paper can be freely obtained from the World Wide Web ( References [1] Hamid R. Berenji and Pratap Khedkar. Learning and tuning fuzzy logic controllers through reinforcements. IEEE Trans. Neural Networks, 3:724{740, September [2] I. Kononenko. On biases in estimating multi-valued attributes. In Proc. 1st International Conference on Knowledge Discovery and Data Mining, pages 1034{1040, [3] Chin-Teng Lin. Neural Fuzzy Control Systems with Structure and Parameter Learning. World Scientic, Singapore, [4] C.J. Merz and P.M. Murphy. UCI repository of machine learning databases. University of California, Irvine, Dept. of Information and Computer Sciences, URL: [5] Detlef Nauck, Frank Klawonn, and Rudolf Kruse. Foundations of Neuro-Fuzzy Systems. Wiley, Chichester, [6] Detlef Nauck and Rudolf Kruse. NEFCLASS { a neuro-fuzzy approach for the classication of data. In K. M. George, Janice H. Carrol, Ed Deaton, Dave Oppenheim, and Jim Hightower, editors, Applied Computing Proc ACM Symposium on Applied Computing, Nashville, Feb. 26{28, pages 461{465. ACM Press, New York, February [7] Detlef Nauck and Rudolf Kruse. A neuro-fuzzy method to learn fuzzy classication rules from data. Fuzzy Sets and Systems, 89:277{288, [8] Detlef Nauck and Rudolf Kruse. New learning strategies for NEFCLASS. In Proc. Seventh International Fuzzy Systems Association World Congress IFSA'97, volume IV, pages 50{55, Prague, [9] J.R. Quinlan. Induction of decision trees. Machine Learning, 1:81{106, [10] J.R. Quinlan. C4.5: Programs for Machine Learning. Morgan Kaufman, San Mateo, CA, [11] J. Rissanen. A universal prior for integers and estimation by minimum description length. Annals of Statistics, 11:416{431, [12] Nadine Tschichold-Gurman. Generation and improvement of fuzzy classiers with incremental learning using fuzzy rulenet. In K. M. George, Janice H. Carrol, Ed Deaton, Dave Oppenheim, and Jim Hightower, editors, Applied Computing Proc ACM Symposium on Applied Computing, Nashville, Feb. 26{28, pages 466{470. ACM Press, New York, February [13] W.H. Wolberg and O.L. Mangasarian. Multisurface method of pattern separation for medical diagnosis applied to breast cytology. Proc. National Academy of Sciences, 87:9193{9196, December 1990.

Some neuro-fuzzy models use more than 3 layers, and encode fuzzy sets as activation functions. In this case, it is usually possible to transform them

Some neuro-fuzzy models use more than 3 layers, and encode fuzzy sets as activation functions. In this case, it is usually possible to transform them NEURO-FUZZY SYSTEMS: REVIEW AND PROSPECTS Detlef Nauck Faculty of Computer Science (FIN-IIK), University of Magdeburg Universitaetsplatz, D-39106 Magdeburg, Germany Tel. +49.391.67-12700, Fax: +49.391.67-12018

More information

APPLICATION OF THE FUZZY MIN-MAX NEURAL NETWORK CLASSIFIER TO PROBLEMS WITH CONTINUOUS AND DISCRETE ATTRIBUTES

APPLICATION OF THE FUZZY MIN-MAX NEURAL NETWORK CLASSIFIER TO PROBLEMS WITH CONTINUOUS AND DISCRETE ATTRIBUTES APPLICATION OF THE FUZZY MIN-MAX NEURAL NETWORK CLASSIFIER TO PROBLEMS WITH CONTINUOUS AND DISCRETE ATTRIBUTES A. Likas, K. Blekas and A. Stafylopatis National Technical University of Athens Department

More information

NEFCLASS A NEURO{FUZZY APPROACH FOR THE CLASSIFICATION OF DATA. Detlef Nauck and Rudolf Kruse. The fuzzy rules describing the data are of the form:

NEFCLASS A NEURO{FUZZY APPROACH FOR THE CLASSIFICATION OF DATA. Detlef Nauck and Rudolf Kruse. The fuzzy rules describing the data are of the form: NEFCLASS A NEURO{FUZZY APPROACH FOR THE CLASSIFICATION OF DATA Detlef Nauck and Rudolf Kruse Technical University of Braunschweig, Dept. of Computer Science Keywords Data analysis, learning, fuzzy system,

More information

volved in the system are usually not dierentiable. The term \fuzzy error" denotes that the error measure guiding the learning process is either specie

volved in the system are usually not dierentiable. The term \fuzzy error denotes that the error measure guiding the learning process is either specie A Neuro-Fuzzy Approach to Obtain Interpretable Fuzzy Systems for Function Approximation Detlef Nauck and Rudolf Kruse University of Magdeburg, Faculty of Computer Science, Universitaetsplatz 2, D-39106

More information

Fuzzy Partitioning with FID3.1

Fuzzy Partitioning with FID3.1 Fuzzy Partitioning with FID3.1 Cezary Z. Janikow Dept. of Mathematics and Computer Science University of Missouri St. Louis St. Louis, Missouri 63121 janikow@umsl.edu Maciej Fajfer Institute of Computing

More information

Evaluating the SVM Component in Oracle 10g Beta

Evaluating the SVM Component in Oracle 10g Beta Evaluating the SVM Component in Oracle 10g Beta Dept. of Computer Science and Statistics University of Rhode Island Technical Report TR04-299 Lutz Hamel and Angela Uvarov Department of Computer Science

More information

A Classifier with the Function-based Decision Tree

A Classifier with the Function-based Decision Tree A Classifier with the Function-based Decision Tree Been-Chian Chien and Jung-Yi Lin Institute of Information Engineering I-Shou University, Kaohsiung 84008, Taiwan, R.O.C E-mail: cbc@isu.edu.tw, m893310m@isu.edu.tw

More information

Design of Neuro Fuzzy Systems

Design of Neuro Fuzzy Systems International Journal of Engineering Research and Technology. ISSN 0974-3154 Volume 6, Number 5 (2013), pp. 695-700 International Research Publication House http://www.irphouse.com Design of Neuro Fuzzy

More information

size, runs an existing induction algorithm on the rst subset to obtain a rst set of rules, and then processes each of the remaining data subsets at a

size, runs an existing induction algorithm on the rst subset to obtain a rst set of rules, and then processes each of the remaining data subsets at a Multi-Layer Incremental Induction Xindong Wu and William H.W. Lo School of Computer Science and Software Ebgineering Monash University 900 Dandenong Road Melbourne, VIC 3145, Australia Email: xindong@computer.org

More information

COMP 465: Data Mining Classification Basics

COMP 465: Data Mining Classification Basics Supervised vs. Unsupervised Learning COMP 465: Data Mining Classification Basics Slides Adapted From : Jiawei Han, Micheline Kamber & Jian Pei Data Mining: Concepts and Techniques, 3 rd ed. Supervised

More information

Minoru SASAKI and Kenji KITA. Department of Information Science & Intelligent Systems. Faculty of Engineering, Tokushima University

Minoru SASAKI and Kenji KITA. Department of Information Science & Intelligent Systems. Faculty of Engineering, Tokushima University Information Retrieval System Using Concept Projection Based on PDDP algorithm Minoru SASAKI and Kenji KITA Department of Information Science & Intelligent Systems Faculty of Engineering, Tokushima University

More information

Look-Ahead Based Fuzzy Decision Tree Induction

Look-Ahead Based Fuzzy Decision Tree Induction IEEE TRANSACTIONS ON FUZZY SYSTEMS, VOL. 9, NO. 3, JUNE 2001 461 Look-Ahead Based Fuzzy Decision Tree Induction Ming Dong, Student Member, IEEE, and Ravi Kothari, Senior Member, IEEE Abstract Decision

More information

An Information-Theoretic Approach to the Prepruning of Classification Rules

An Information-Theoretic Approach to the Prepruning of Classification Rules An Information-Theoretic Approach to the Prepruning of Classification Rules Max Bramer University of Portsmouth, Portsmouth, UK Abstract: Keywords: The automatic induction of classification rules from

More information

Networks for Control. California Institute of Technology. Pasadena, CA Abstract

Networks for Control. California Institute of Technology. Pasadena, CA Abstract Learning Fuzzy Rule-Based Neural Networks for Control Charles M. Higgins and Rodney M. Goodman Department of Electrical Engineering, 116-81 California Institute of Technology Pasadena, CA 91125 Abstract

More information

HARD, SOFT AND FUZZY C-MEANS CLUSTERING TECHNIQUES FOR TEXT CLASSIFICATION

HARD, SOFT AND FUZZY C-MEANS CLUSTERING TECHNIQUES FOR TEXT CLASSIFICATION HARD, SOFT AND FUZZY C-MEANS CLUSTERING TECHNIQUES FOR TEXT CLASSIFICATION 1 M.S.Rekha, 2 S.G.Nawaz 1 PG SCALOR, CSE, SRI KRISHNADEVARAYA ENGINEERING COLLEGE, GOOTY 2 ASSOCIATE PROFESSOR, SRI KRISHNADEVARAYA

More information

Han Liu, Alexander Gegov & Mihaela Cocea

Han Liu, Alexander Gegov & Mihaela Cocea Rule-based systems: a granular computing perspective Han Liu, Alexander Gegov & Mihaela Cocea Granular Computing ISSN 2364-4966 Granul. Comput. DOI 10.1007/s41066-016-0021-6 1 23 Your article is published

More information

Improving Tree-Based Classification Rules Using a Particle Swarm Optimization

Improving Tree-Based Classification Rules Using a Particle Swarm Optimization Improving Tree-Based Classification Rules Using a Particle Swarm Optimization Chi-Hyuck Jun *, Yun-Ju Cho, and Hyeseon Lee Department of Industrial and Management Engineering Pohang University of Science

More information

Feature-weighted k-nearest Neighbor Classifier

Feature-weighted k-nearest Neighbor Classifier Proceedings of the 27 IEEE Symposium on Foundations of Computational Intelligence (FOCI 27) Feature-weighted k-nearest Neighbor Classifier Diego P. Vivencio vivencio@comp.uf scar.br Estevam R. Hruschka

More information

Association Rule Mining and Clustering

Association Rule Mining and Clustering Association Rule Mining and Clustering Lecture Outline: Classification vs. Association Rule Mining vs. Clustering Association Rule Mining Clustering Types of Clusters Clustering Algorithms Hierarchical:

More information

FEATURE EXTRACTION USING FUZZY RULE BASED SYSTEM

FEATURE EXTRACTION USING FUZZY RULE BASED SYSTEM International Journal of Computer Science and Applications, Vol. 5, No. 3, pp 1-8 Technomathematics Research Foundation FEATURE EXTRACTION USING FUZZY RULE BASED SYSTEM NARENDRA S. CHAUDHARI and AVISHEK

More information

Rowena Cole and Luigi Barone. Department of Computer Science, The University of Western Australia, Western Australia, 6907

Rowena Cole and Luigi Barone. Department of Computer Science, The University of Western Australia, Western Australia, 6907 The Game of Clustering Rowena Cole and Luigi Barone Department of Computer Science, The University of Western Australia, Western Australia, 697 frowena, luigig@cs.uwa.edu.au Abstract Clustering is a technique

More information

CloNI: clustering of JN -interval discretization

CloNI: clustering of JN -interval discretization CloNI: clustering of JN -interval discretization C. Ratanamahatana Department of Computer Science, University of California, Riverside, USA Abstract It is known that the naive Bayesian classifier typically

More information

Improving Surface Defect Detection for Quality Assessment of Car Body Panels

Improving Surface Defect Detection for Quality Assessment of Car Body Panels Mathware & Soft Computing 11 (2004) 163-177 Improving Surface Defect Detection for Quality Assessment of Car Body Panels C. Döring +, A. Eichhorn, D. Girimonte and R. Kruse + + University of Magdeburg,

More information

Comparative Study of Clustering Algorithms using R

Comparative Study of Clustering Algorithms using R Comparative Study of Clustering Algorithms using R Debayan Das 1 and D. Peter Augustine 2 1 ( M.Sc Computer Science Student, Christ University, Bangalore, India) 2 (Associate Professor, Department of Computer

More information

Unsupervised Learning

Unsupervised Learning Outline Unsupervised Learning Basic concepts K-means algorithm Representation of clusters Hierarchical clustering Distance functions Which clustering algorithm to use? NN Supervised learning vs. unsupervised

More information

2 Data Reduction Techniques The granularity of reducible information is one of the main criteria for classifying the reduction techniques. While the t

2 Data Reduction Techniques The granularity of reducible information is one of the main criteria for classifying the reduction techniques. While the t Data Reduction - an Adaptation Technique for Mobile Environments A. Heuer, A. Lubinski Computer Science Dept., University of Rostock, Germany Keywords. Reduction. Mobile Database Systems, Data Abstract.

More information

Analyzing Borders Between Partially Contradicting Fuzzy Classification Rules

Analyzing Borders Between Partially Contradicting Fuzzy Classification Rules Analyzing Borders Between Partially Contradicting Fuzzy Classification Rules Andreas Nürnberger, Aljoscha Klose, Rudolf Kruse University of Magdeburg, Faculty of Comuter Science 39106 Magdeburg, Germany

More information

Induction of Association Rules: Apriori Implementation

Induction of Association Rules: Apriori Implementation 1 Induction of Association Rules: Apriori Implementation Christian Borgelt and Rudolf Kruse Department of Knowledge Processing and Language Engineering School of Computer Science Otto-von-Guericke-University

More information

GENERATING FUZZY RULES FROM EXAMPLES USING GENETIC. Francisco HERRERA, Manuel LOZANO, Jose Luis VERDEGAY

GENERATING FUZZY RULES FROM EXAMPLES USING GENETIC. Francisco HERRERA, Manuel LOZANO, Jose Luis VERDEGAY GENERATING FUZZY RULES FROM EXAMPLES USING GENETIC ALGORITHMS Francisco HERRERA, Manuel LOZANO, Jose Luis VERDEGAY Dept. of Computer Science and Articial Intelligence University of Granada, 18071 - Granada,

More information

Toward Automated Cancer Diagnosis: An Interactive System for Cell Feature Extraction

Toward Automated Cancer Diagnosis: An Interactive System for Cell Feature Extraction Toward Automated Cancer Diagnosis: An Interactive System for Cell Feature Extraction Nick Street Computer Sciences Department University of Wisconsin-Madison street@cs.wisc.edu Abstract Oncologists at

More information

Estimating Missing Attribute Values Using Dynamically-Ordered Attribute Trees

Estimating Missing Attribute Values Using Dynamically-Ordered Attribute Trees Estimating Missing Attribute Values Using Dynamically-Ordered Attribute Trees Jing Wang Computer Science Department, The University of Iowa jing-wang-1@uiowa.edu W. Nick Street Management Sciences Department,

More information

Clustering Sequences with Hidden. Markov Models. Padhraic Smyth CA Abstract

Clustering Sequences with Hidden. Markov Models. Padhraic Smyth CA Abstract Clustering Sequences with Hidden Markov Models Padhraic Smyth Information and Computer Science University of California, Irvine CA 92697-3425 smyth@ics.uci.edu Abstract This paper discusses a probabilistic

More information

Genetic Tuning for Improving Wang and Mendel s Fuzzy Database

Genetic Tuning for Improving Wang and Mendel s Fuzzy Database Proceedings of the 2009 IEEE International Conference on Systems, Man, and Cybernetics San Antonio, TX, USA - October 2009 Genetic Tuning for Improving Wang and Mendel s Fuzzy Database E. R. R. Kato, O.

More information

Processing Missing Values with Self-Organized Maps

Processing Missing Values with Self-Organized Maps Processing Missing Values with Self-Organized Maps David Sommer, Tobias Grimm, Martin Golz University of Applied Sciences Schmalkalden Department of Computer Science D-98574 Schmalkalden, Germany Phone:

More information

Data Mining: Concepts and Techniques Classification and Prediction Chapter 6.1-3

Data Mining: Concepts and Techniques Classification and Prediction Chapter 6.1-3 Data Mining: Concepts and Techniques Classification and Prediction Chapter 6.1-3 January 25, 2007 CSE-4412: Data Mining 1 Chapter 6 Classification and Prediction 1. What is classification? What is prediction?

More information

Identifying fuzzy systems from numerical data with Xfuzzy

Identifying fuzzy systems from numerical data with Xfuzzy Identifying fuzzy systems from numerical data with Xfuzzy I. Baturone Instituto de Microelectrónica de Sevilla, (IMSE-CNM), y Dept. Electrónica y Electromagnetismo, Univ. de Sevilla (Spain) lumi@imse.cnm.es

More information

Analysis of Decision Boundaries Generated by Constructive Neural Network Learning Algorithms

Analysis of Decision Boundaries Generated by Constructive Neural Network Learning Algorithms Computer Science Technical Reports Computer Science 995 Analysis of Decision Boundaries Generated by Constructive Neural Network Learning Algorithms ChunHsien Chen Iowa State University R. G. Parekh Iowa

More information

Fuzzy data analysis with NEFCLASS

Fuzzy data analysis with NEFCLASS International Journal of Approximate Reasoning 32(2003) 103 130 www.elsevier.com/locate/ijar Fuzzy data analysis with NEFCLASS Detlef D. Nauck * BTexact Technologies, Intelligent Systems Lab, Orion Building

More information

A New Method For Forecasting Enrolments Combining Time-Variant Fuzzy Logical Relationship Groups And K-Means Clustering

A New Method For Forecasting Enrolments Combining Time-Variant Fuzzy Logical Relationship Groups And K-Means Clustering A New Method For Forecasting Enrolments Combining Time-Variant Fuzzy Logical Relationship Groups And K-Means Clustering Nghiem Van Tinh 1, Vu Viet Vu 1, Tran Thi Ngoc Linh 1 1 Thai Nguyen University of

More information

Lab 2: Support Vector Machines

Lab 2: Support Vector Machines Articial neural networks, advanced course, 2D1433 Lab 2: Support Vector Machines March 13, 2007 1 Background Support vector machines, when used for classication, nd a hyperplane w, x + b = 0 that separates

More information

r=0 for instances discarded at A r=1 for instances discarded at B r=2 for instances discarded at C r=3 for instances at the leaf node

r=0 for instances discarded at A r=1 for instances discarded at B r=2 for instances discarded at C r=3 for instances at the leaf node Boosting Lazy Decision Trees Xiaoli Zhang Fern xz@ecn.purdue.edu Carla E. Brodley brodley@ecn.purdue.edu School of Electrical and Computer Engineering, Purdue University, West Lafayette, IN 47907 Abstract

More information

Multiobjective Formulations of Fuzzy Rule-Based Classification System Design

Multiobjective Formulations of Fuzzy Rule-Based Classification System Design Multiobjective Formulations of Fuzzy Rule-Based Classification System Design Hisao Ishibuchi and Yusuke Nojima Graduate School of Engineering, Osaka Prefecture University, - Gakuen-cho, Sakai, Osaka 599-853,

More information

Classifier C-Net. 2D Projected Images of 3D Objects. 2D Projected Images of 3D Objects. Model I. Model II

Classifier C-Net. 2D Projected Images of 3D Objects. 2D Projected Images of 3D Objects. Model I. Model II Advances in Neural Information Processing Systems 7. (99) The MIT Press, Cambridge, MA. pp.949-96 Unsupervised Classication of 3D Objects from D Views Satoshi Suzuki Hiroshi Ando ATR Human Information

More information

A Systematic Overview of Data Mining Algorithms. Sargur Srihari University at Buffalo The State University of New York

A Systematic Overview of Data Mining Algorithms. Sargur Srihari University at Buffalo The State University of New York A Systematic Overview of Data Mining Algorithms Sargur Srihari University at Buffalo The State University of New York 1 Topics Data Mining Algorithm Definition Example of CART Classification Iris, Wine

More information

Evolving SQL Queries for Data Mining

Evolving SQL Queries for Data Mining Evolving SQL Queries for Data Mining Majid Salim and Xin Yao School of Computer Science, The University of Birmingham Edgbaston, Birmingham B15 2TT, UK {msc30mms,x.yao}@cs.bham.ac.uk Abstract. This paper

More information

Logistic Model Tree With Modified AIC

Logistic Model Tree With Modified AIC Logistic Model Tree With Modified AIC Mitesh J. Thakkar Neha J. Thakkar Dr. J.S.Shah Student of M.E.I.T. Asst.Prof.Computer Dept. Prof.&Head Computer Dept. S.S.Engineering College, Indus Engineering College

More information

Fuzzy Data Analysis: Challenges and Perspectives

Fuzzy Data Analysis: Challenges and Perspectives Fuzzy Data Analysis: Challenges and Perspectives Rudolf Kruse, Christian Borgelt, and Detlef Nauck Dept. of Knowledge Processing and Language Engineering Otto-von-Guericke-University of Magdeburg D-39106

More information

Resolving the Conflict Between Competitive and Cooperative Behavior in Michigan-Type Fuzzy Classifier Systems

Resolving the Conflict Between Competitive and Cooperative Behavior in Michigan-Type Fuzzy Classifier Systems Resolving the Conflict Between Competitive and Cooperative Behavior in Michigan-Type Fuzzy Classifier Systems Peter Haslinger and Ulrich Bodenhofer Software Competence Center Hagenberg A-4232 Hagenberg,

More information

Using Local Trajectory Optimizers To Speed Up Global. Christopher G. Atkeson. Department of Brain and Cognitive Sciences and

Using Local Trajectory Optimizers To Speed Up Global. Christopher G. Atkeson. Department of Brain and Cognitive Sciences and Using Local Trajectory Optimizers To Speed Up Global Optimization In Dynamic Programming Christopher G. Atkeson Department of Brain and Cognitive Sciences and the Articial Intelligence Laboratory Massachusetts

More information

Challenges and Interesting Research Directions in Associative Classification

Challenges and Interesting Research Directions in Associative Classification Challenges and Interesting Research Directions in Associative Classification Fadi Thabtah Department of Management Information Systems Philadelphia University Amman, Jordan Email: FFayez@philadelphia.edu.jo

More information

A taxonomy of race. D. P. Helmbold, C. E. McDowell. September 28, University of California, Santa Cruz. Santa Cruz, CA

A taxonomy of race. D. P. Helmbold, C. E. McDowell. September 28, University of California, Santa Cruz. Santa Cruz, CA A taxonomy of race conditions. D. P. Helmbold, C. E. McDowell UCSC-CRL-94-34 September 28, 1994 Board of Studies in Computer and Information Sciences University of California, Santa Cruz Santa Cruz, CA

More information

Concept Tree Based Clustering Visualization with Shaded Similarity Matrices

Concept Tree Based Clustering Visualization with Shaded Similarity Matrices Syracuse University SURFACE School of Information Studies: Faculty Scholarship School of Information Studies (ischool) 12-2002 Concept Tree Based Clustering Visualization with Shaded Similarity Matrices

More information

Contrained K-Means Clustering 1 1 Introduction The K-Means clustering algorithm [5] has become a workhorse for the data analyst in many diverse elds.

Contrained K-Means Clustering 1 1 Introduction The K-Means clustering algorithm [5] has become a workhorse for the data analyst in many diverse elds. Constrained K-Means Clustering P. S. Bradley K. P. Bennett A. Demiriz Microsoft Research Dept. of Mathematical Sciences One Microsoft Way Dept. of Decision Sciences and Eng. Sys. Redmond, WA 98052 Renselaer

More information

CHAPTER 4 FREQUENCY STABILIZATION USING FUZZY LOGIC CONTROLLER

CHAPTER 4 FREQUENCY STABILIZATION USING FUZZY LOGIC CONTROLLER 60 CHAPTER 4 FREQUENCY STABILIZATION USING FUZZY LOGIC CONTROLLER 4.1 INTRODUCTION Problems in the real world quite often turn out to be complex owing to an element of uncertainty either in the parameters

More information

Performance Analysis of Data Mining Classification Techniques

Performance Analysis of Data Mining Classification Techniques Performance Analysis of Data Mining Classification Techniques Tejas Mehta 1, Dr. Dhaval Kathiriya 2 Ph.D. Student, School of Computer Science, Dr. Babasaheb Ambedkar Open University, Gujarat, India 1 Principal

More information

Classification Using Unstructured Rules and Ant Colony Optimization

Classification Using Unstructured Rules and Ant Colony Optimization Classification Using Unstructured Rules and Ant Colony Optimization Negar Zakeri Nejad, Amir H. Bakhtiary, and Morteza Analoui Abstract In this paper a new method based on the algorithm is proposed to

More information

BRACE: A Paradigm For the Discretization of Continuously Valued Data

BRACE: A Paradigm For the Discretization of Continuously Valued Data Proceedings of the Seventh Florida Artificial Intelligence Research Symposium, pp. 7-2, 994 BRACE: A Paradigm For the Discretization of Continuously Valued Data Dan Ventura Tony R. Martinez Computer Science

More information

LEARNING WEIGHTS OF FUZZY RULES BY USING GRAVITATIONAL SEARCH ALGORITHM

LEARNING WEIGHTS OF FUZZY RULES BY USING GRAVITATIONAL SEARCH ALGORITHM International Journal of Innovative Computing, Information and Control ICIC International c 2013 ISSN 1349-4198 Volume 9, Number 4, April 2013 pp. 1593 1601 LEARNING WEIGHTS OF FUZZY RULES BY USING GRAVITATIONAL

More information

Fuzzy Reasoning. Linguistic Variables

Fuzzy Reasoning. Linguistic Variables Fuzzy Reasoning Linguistic Variables Linguistic variable is an important concept in fuzzy logic and plays a key role in its applications, especially in the fuzzy expert system Linguistic variable is a

More information

GEOG 5113 Special Topics in GIScience. Why is Classical set theory restricted? Contradiction & Excluded Middle. Fuzzy Set Theory in GIScience

GEOG 5113 Special Topics in GIScience. Why is Classical set theory restricted? Contradiction & Excluded Middle. Fuzzy Set Theory in GIScience GEOG 5113 Special Topics in GIScience Fuzzy Set Theory in GIScience -Basic Properties and Concepts of Fuzzy Sets- Why is Classical set theory restricted? Boundaries of classical sets are required to be

More information

Egemen Tanin, Tahsin M. Kurc, Cevdet Aykanat, Bulent Ozguc. Abstract. Direct Volume Rendering (DVR) is a powerful technique for

Egemen Tanin, Tahsin M. Kurc, Cevdet Aykanat, Bulent Ozguc. Abstract. Direct Volume Rendering (DVR) is a powerful technique for Comparison of Two Image-Space Subdivision Algorithms for Direct Volume Rendering on Distributed-Memory Multicomputers Egemen Tanin, Tahsin M. Kurc, Cevdet Aykanat, Bulent Ozguc Dept. of Computer Eng. and

More information

MISSING VALUES AND LEARNING OF FUZZY RULES MICHAEL R. BERTHOLD. Computer Science Division, Department of EECS

MISSING VALUES AND LEARNING OF FUZZY RULES MICHAEL R. BERTHOLD. Computer Science Division, Department of EECS MISSING VALUES AND LEARNING OF FUZZY RULES MICHAEL R. BERTHOLD Computer Science Division, Department of EECS University of California, Bereley, CA 94720, USA email: Michael.Berthold@Informati.Uni-Karlsruhe.DE

More information

Speeding Up Fuzzy Clustering with Neural Network Techniques

Speeding Up Fuzzy Clustering with Neural Network Techniques Speeding Up Fuzzy Clustering with Neural Network Techniques Christian Borgelt and Rudolf Kruse Research Group Neural Networks and Fuzzy Systems Dept. of Knowledge Processing and Language Engineering, School

More information

Extra readings beyond the lecture slides are important:

Extra readings beyond the lecture slides are important: 1 Notes To preview next lecture: Check the lecture notes, if slides are not available: http://web.cse.ohio-state.edu/~sun.397/courses/au2017/cse5243-new.html Check UIUC course on the same topic. All their

More information

Using Decision Boundary to Analyze Classifiers

Using Decision Boundary to Analyze Classifiers Using Decision Boundary to Analyze Classifiers Zhiyong Yan Congfu Xu College of Computer Science, Zhejiang University, Hangzhou, China yanzhiyong@zju.edu.cn Abstract In this paper we propose to use decision

More information

Efficient Tuning of SVM Hyperparameters Using Radius/Margin Bound and Iterative Algorithms

Efficient Tuning of SVM Hyperparameters Using Radius/Margin Bound and Iterative Algorithms IEEE TRANSACTIONS ON NEURAL NETWORKS, VOL. 13, NO. 5, SEPTEMBER 2002 1225 Efficient Tuning of SVM Hyperparameters Using Radius/Margin Bound and Iterative Algorithms S. Sathiya Keerthi Abstract This paper

More information

Keywords: clustering algorithms, unsupervised learning, cluster validity

Keywords: clustering algorithms, unsupervised learning, cluster validity Volume 6, Issue 1, January 2016 ISSN: 2277 128X International Journal of Advanced Research in Computer Science and Software Engineering Research Paper Available online at: www.ijarcsse.com Clustering Based

More information

Application of fuzzy set theory in image analysis. Nataša Sladoje Centre for Image Analysis

Application of fuzzy set theory in image analysis. Nataša Sladoje Centre for Image Analysis Application of fuzzy set theory in image analysis Nataša Sladoje Centre for Image Analysis Our topics for today Crisp vs fuzzy Fuzzy sets and fuzzy membership functions Fuzzy set operators Approximate

More information

The task of inductive learning from examples is to nd an approximate definition

The task of inductive learning from examples is to nd an approximate definition 1 Initializing Neural Networks using Decision Trees Arunava Banerjee 1.1 Introduction The task of inductive learning from examples is to nd an approximate definition for an unknown function f(x), given

More information

Flexible-Hybrid Sequential Floating Search in Statistical Feature Selection

Flexible-Hybrid Sequential Floating Search in Statistical Feature Selection Flexible-Hybrid Sequential Floating Search in Statistical Feature Selection Petr Somol 1,2, Jana Novovičová 1,2, and Pavel Pudil 2,1 1 Dept. of Pattern Recognition, Institute of Information Theory and

More information

Logical Decision Rules: Teaching C4.5 to Speak Prolog

Logical Decision Rules: Teaching C4.5 to Speak Prolog Logical Decision Rules: Teaching C4.5 to Speak Prolog Kamran Karimi and Howard J. Hamilton Department of Computer Science University of Regina Regina, Saskatchewan Canada S4S 0A2 {karimi,hamilton}@cs.uregina.ca

More information

FuzzyDT- A Fuzzy Decision Tree Algorithm Based on C4.5

FuzzyDT- A Fuzzy Decision Tree Algorithm Based on C4.5 FuzzyDT- A Fuzzy Decision Tree Algorithm Based on C4.5 Marcos E. Cintra 1, Maria C. Monard 2, and Heloisa A. Camargo 3 1 Exact and Natural Sciences Dept. - Federal University of the Semi-arid - UFERSA

More information

AST: Support for Algorithm Selection with a CBR Approach

AST: Support for Algorithm Selection with a CBR Approach AST: Support for Algorithm Selection with a CBR Approach Guido Lindner 1 and Rudi Studer 2 1 DaimlerChrysler AG, Research &Technology FT3/KL, PO: DaimlerChrysler AG, T-402, D-70456 Stuttgart, Germany guido.lindner@daimlerchrysler.com

More information

Cluster quality 15. Running time 0.7. Distance between estimated and true means Running time [s]

Cluster quality 15. Running time 0.7. Distance between estimated and true means Running time [s] Fast, single-pass K-means algorithms Fredrik Farnstrom Computer Science and Engineering Lund Institute of Technology, Sweden arnstrom@ucsd.edu James Lewis Computer Science and Engineering University of

More information

Fuzzy If-Then Rules. Fuzzy If-Then Rules. Adnan Yazıcı

Fuzzy If-Then Rules. Fuzzy If-Then Rules. Adnan Yazıcı Fuzzy If-Then Rules Adnan Yazıcı Dept. of Computer Engineering, Middle East Technical University Ankara/Turkey Fuzzy If-Then Rules There are two different kinds of fuzzy rules: Fuzzy mapping rules and

More information

6. Dicretization methods 6.1 The purpose of discretization

6. Dicretization methods 6.1 The purpose of discretization 6. Dicretization methods 6.1 The purpose of discretization Often data are given in the form of continuous values. If their number is huge, model building for such data can be difficult. Moreover, many

More information

Huan Liu. Kent Ridge, Singapore Tel: (+65) ; Fax: (+65) Abstract

Huan Liu. Kent Ridge, Singapore Tel: (+65) ; Fax: (+65) Abstract A Family of Ecient Rule Generators Huan Liu Department of Information Systems and Computer Science National University of Singapore Kent Ridge, Singapore 119260 liuh@iscs.nus.sg Tel: (+65)-772-6563; Fax:

More information

OBJECT-CENTERED INTERACTIVE MULTI-DIMENSIONAL SCALING: ASK THE EXPERT

OBJECT-CENTERED INTERACTIVE MULTI-DIMENSIONAL SCALING: ASK THE EXPERT OBJECT-CENTERED INTERACTIVE MULTI-DIMENSIONAL SCALING: ASK THE EXPERT Joost Broekens Tim Cocx Walter A. Kosters Leiden Institute of Advanced Computer Science Leiden University, The Netherlands Email: {broekens,

More information

FUZZY BOOLEAN ALGEBRAS AND LUKASIEWICZ LOGIC. Angel Garrido

FUZZY BOOLEAN ALGEBRAS AND LUKASIEWICZ LOGIC. Angel Garrido Acta Universitatis Apulensis ISSN: 1582-5329 No. 22/2010 pp. 101-111 FUZZY BOOLEAN ALGEBRAS AND LUKASIEWICZ LOGIC Angel Garrido Abstract. In this paper, we analyze the more adequate tools to solve many

More information

Segmentation of Images

Segmentation of Images Segmentation of Images SEGMENTATION If an image has been preprocessed appropriately to remove noise and artifacts, segmentation is often the key step in interpreting the image. Image segmentation is a

More information

Semi-Supervised Clustering with Partial Background Information

Semi-Supervised Clustering with Partial Background Information Semi-Supervised Clustering with Partial Background Information Jing Gao Pang-Ning Tan Haibin Cheng Abstract Incorporating background knowledge into unsupervised clustering algorithms has been the subject

More information

Improving the Wang and Mendel s Fuzzy Rule Learning Method by Inducing Cooperation Among Rules 1

Improving the Wang and Mendel s Fuzzy Rule Learning Method by Inducing Cooperation Among Rules 1 Improving the Wang and Mendel s Fuzzy Rule Learning Method by Inducing Cooperation Among Rules 1 J. Casillas DECSAI, University of Granada 18071 Granada, Spain casillas@decsai.ugr.es O. Cordón DECSAI,

More information

C-NBC: Neighborhood-Based Clustering with Constraints

C-NBC: Neighborhood-Based Clustering with Constraints C-NBC: Neighborhood-Based Clustering with Constraints Piotr Lasek Chair of Computer Science, University of Rzeszów ul. Prof. St. Pigonia 1, 35-310 Rzeszów, Poland lasek@ur.edu.pl Abstract. Clustering is

More information

the number of states must be set in advance, i.e. the structure of the model is not t to the data, but given a priori the algorithm converges to a loc

the number of states must be set in advance, i.e. the structure of the model is not t to the data, but given a priori the algorithm converges to a loc Clustering Time Series with Hidden Markov Models and Dynamic Time Warping Tim Oates, Laura Firoiu and Paul R. Cohen Computer Science Department, LGRC University of Massachusetts, Box 34610 Amherst, MA

More information

Automatic Generation of Fuzzy Classification Rules Using Granulation-Based Adaptive Clustering

Automatic Generation of Fuzzy Classification Rules Using Granulation-Based Adaptive Clustering Automatic Generation of Fuzzy Classification Rules Using Granulation-Based Adaptive Clustering Mohammed Al-Shammaa*, Maysam F. Abbod Department of Electronic and Computer Engineering Brunel University

More information

Dynamic Clustering of Data with Modified K-Means Algorithm

Dynamic Clustering of Data with Modified K-Means Algorithm 2012 International Conference on Information and Computer Networks (ICICN 2012) IPCSIT vol. 27 (2012) (2012) IACSIT Press, Singapore Dynamic Clustering of Data with Modified K-Means Algorithm Ahamed Shafeeq

More information

(Preliminary Version 2 ) Jai-Hoon Kim Nitin H. Vaidya. Department of Computer Science. Texas A&M University. College Station, TX

(Preliminary Version 2 ) Jai-Hoon Kim Nitin H. Vaidya. Department of Computer Science. Texas A&M University. College Station, TX Towards an Adaptive Distributed Shared Memory (Preliminary Version ) Jai-Hoon Kim Nitin H. Vaidya Department of Computer Science Texas A&M University College Station, TX 77843-3 E-mail: fjhkim,vaidyag@cs.tamu.edu

More information

7. Decision or classification trees

7. Decision or classification trees 7. Decision or classification trees Next we are going to consider a rather different approach from those presented so far to machine learning that use one of the most common and important data structure,

More information

Recent Progress on RAIL: Automating Clustering and Comparison of Different Road Classification Techniques on High Resolution Remotely Sensed Imagery

Recent Progress on RAIL: Automating Clustering and Comparison of Different Road Classification Techniques on High Resolution Remotely Sensed Imagery Recent Progress on RAIL: Automating Clustering and Comparison of Different Road Classification Techniques on High Resolution Remotely Sensed Imagery Annie Chen ANNIEC@CSE.UNSW.EDU.AU Gary Donovan GARYD@CSE.UNSW.EDU.AU

More information

An On-line Variable Length Binary. Institute for Systems Research and. Institute for Advanced Computer Studies. University of Maryland

An On-line Variable Length Binary. Institute for Systems Research and. Institute for Advanced Computer Studies. University of Maryland An On-line Variable Length inary Encoding Tinku Acharya Joseph F. Ja Ja Institute for Systems Research and Institute for Advanced Computer Studies University of Maryland College Park, MD 242 facharya,

More information

Performance Measures for Multi-Graded Relevance

Performance Measures for Multi-Graded Relevance Performance Measures for Multi-Graded Relevance Christian Scheel, Andreas Lommatzsch, and Sahin Albayrak Technische Universität Berlin, DAI-Labor, Germany {christian.scheel,andreas.lommatzsch,sahin.albayrak}@dai-labor.de

More information

Accelerating Unique Strategy for Centroid Priming in K-Means Clustering

Accelerating Unique Strategy for Centroid Priming in K-Means Clustering IJIRST International Journal for Innovative Research in Science & Technology Volume 3 Issue 07 December 2016 ISSN (online): 2349-6010 Accelerating Unique Strategy for Centroid Priming in K-Means Clustering

More information

Mass Classification Method in Mammogram Using Fuzzy K-Nearest Neighbour Equality

Mass Classification Method in Mammogram Using Fuzzy K-Nearest Neighbour Equality Mass Classification Method in Mammogram Using Fuzzy K-Nearest Neighbour Equality Abstract: Mass classification of objects is an important area of research and application in a variety of fields. In this

More information

Data Mining. 3.2 Decision Tree Classifier. Fall Instructor: Dr. Masoud Yaghini. Chapter 5: Decision Tree Classifier

Data Mining. 3.2 Decision Tree Classifier. Fall Instructor: Dr. Masoud Yaghini. Chapter 5: Decision Tree Classifier Data Mining 3.2 Decision Tree Classifier Fall 2008 Instructor: Dr. Masoud Yaghini Outline Introduction Basic Algorithm for Decision Tree Induction Attribute Selection Measures Information Gain Gain Ratio

More information

Initial Partitions. Region. Region Interpolation. 3. Region parametrization. Ordering. Partition Creation. Interpolated Partitions

Initial Partitions. Region. Region Interpolation. 3. Region parametrization. Ordering. Partition Creation. Interpolated Partitions SEGMENTATION-BASED MORPHOLOGICAL INTERPOLATION OF PARTITION SEQUENCES R. BR EMOND and F. MARQU ES Dept. of Signal Theory and Communications Universitat Politecnica de Catalunya Campus Nord -Modulo D5 C/

More information

Fuzzy Modeling for Control.,,i.

Fuzzy Modeling for Control.,,i. Fuzzy Modeling for Control,,i. INTERNATIONAL SERIES IN INTELLIGENT TECHNOLOGIES Prof. Dr. Dr. h.c. Hans-Jiirgen Zimmermann, Editor European Laboratory for Intelligent Techniques Engineering Aachen, Germany

More information

A Novel Technique for Optimizing the Hidden Layer Architecture in Artificial Neural Networks N. M. Wagarachchi 1, A. S.

A Novel Technique for Optimizing the Hidden Layer Architecture in Artificial Neural Networks N. M. Wagarachchi 1, A. S. American International Journal of Research in Science, Technology, Engineering & Mathematics Available online at http://www.iasir.net ISSN (Print): 2328-3491, ISSN (Online): 2328-3580, ISSN (CD-ROM): 2328-3629

More information

Efficient Pruning Method for Ensemble Self-Generating Neural Networks

Efficient Pruning Method for Ensemble Self-Generating Neural Networks Efficient Pruning Method for Ensemble Self-Generating Neural Networks Hirotaka INOUE Department of Electrical Engineering & Information Science, Kure National College of Technology -- Agaminami, Kure-shi,

More information

Interactive Visualization of Fuzzy Set Operations

Interactive Visualization of Fuzzy Set Operations Interactive Visualization of Fuzzy Set Operations Yeseul Park* a, Jinah Park a a Computer Graphics and Visualization Laboratory, Korea Advanced Institute of Science and Technology, 119 Munji-ro, Yusung-gu,

More information

OC1: A randomized algorithm for building oblique. decision trees. Sreerama K. Murthy Simon Kasif. Steven Salzberg. Department of Computer Science

OC1: A randomized algorithm for building oblique. decision trees. Sreerama K. Murthy Simon Kasif. Steven Salzberg. Department of Computer Science OC1: A randomized algorithm for building oblique decision trees Sreerama K. Murthy Simon Kasif Steven Salzberg Department of Computer Science Johns Hopkins University Baltimore, MD 21218 Richard Beigel

More information