Computational Intelligence for Knowledge%2DBased Systems Design

Information Processing and Management of Uncertainty in Knowledge%2DBased Systems. Theory and Methods %2D 13th International Conference

Information Processing and Management of Uncertainty in Knowledge%2DBased Systems. Applications %2D 13th International Conference

Predicting Partial Orders Ranking with Abstention.

Regret Analysis for Performance Metrics in Multi%2DLabel Classification The Case of Hamming and Subset Zero%2DOne Loss.

Graded Multilabel Classification The Ordinal Case.
We propose a generalization of multilabel classification that we refer to as graded multilabel classification. The key idea is that, instead of requesting a yes-no answer to the question of class membership or, say, relevance of a class label for an instance, we allow for a graded membership of an instance, measured on an ordinal scale of membership degrees. This extension is motivated by practical applications in which a graded or partial class membership is natural. Apart from introducing the basic setting, we propose two general strategies for reducing graded multilabel problems to conventional (multilabel) classification problems. Moreover, we address the question of how to extend performance metrics commonly used in multilabel classification to the graded setting, and present first experimental results. 1.
Bayes Optimal Multilabel Classification via Probabilistic Classifier Chains.
In the realm of multilabel classification (MLC), it has become an opinio communis that optimal predictive performance can only be achieved by learners that explicitly take label dependence into account. The goal of this paper is to elaborate on this postulate in a critical way. To this end, we formalize and analyze MLC within a probabilistic setting. Thus, it becomes possible to look at the problem from the point of view of risk minimization and Bayes optimal prediction. Moreover, inspired by our probabilistic setting, we propose a new method for MLC that generalizes and outperforms another approach, called classifier chains, that was recently introduced in the literature. 1.
Uncertainty in Clustering and Classification.
The KDD process aims at the discovery and extraction of &#034;useful&#034; knowledge (such as interesting patterns, classification, rules etc) from large data repositories. A widely recognized requirement is that the patterns discovered must be valid and ultimately comprehensible (i.e., to be easily understood by analysts). Another requirement that is under-addressed in KDD process is the reveal and the handling of uncertainty in the main data mining processes of clustering, classification and association rules extraction.
An Analysis of the FURIA Algorithm for Fuzzy Rule Induction.

On predictive accuracy and risk minimization in pairwise label ranking.

Combining predictions in pairwise classification An optimal adaptive voting strategy and its relation to weighted voting.

Binary Decomposition Methods for Multipartite Ranking.

Combining Instance%2DBased Learning and Logistic Regression for Multilabel Classification.
Abstract. Multilabel classification is an extension of conventional classification in which a single instance can be associated with multiple labels. Recent research has shown that, just like for standard classification, instance-based learning algorithms relying on the nearest neighbor estimation principle can be used quite successfully in this context. However, since hitherto existing algorithms do not take correlations and interdependencies between labels into account, their potential has not yet been fully exploited. In this paper, we propose a new approach to multilabel classification, which is based on a framework that unifies instancebased learning and logistic regression, comprising both methods as special cases. This approach allows one to capture interdependencies between labels and, moreover, to combine model-based and similarity-based inference for multilabel classification. As will be shown by experimental studies, our approach is able to improve predictive accuracy in terms of several evaluation criteria for multilabel prediction. 1
Graph%2DKernels for the Comparative Analysis of Protein Active Sites.

Decision tree and instance%2Dbased learning for label ranking.
The label ranking problem consists of learning a model that maps instances to total orders over a finite set of predefined labels. This paper introduces new methods for label ranking that complement and improve upon existing approaches. More specifically, we propose extensions of two methods that have been used extensively for classification and regression so far, namely instance-based learning and decision tree induction. The unifying element of the two methods is a procedure for locally estimating predictive probability models for label rankings. 1.
A Fuzzy Variant of the Rand Index for Comparing Clustering Structures.
Abstract — In this paper, we introduce a fuzzy extension of the Rand index, a well-known measure for comparing two clustering structures. In contrast to an existing proposal, which is restricted to the comparison of a fuzzy partition with a non-fuzzy reference partition, our extension is able to compare two proper fuzzy partitions with each other. Elaborating on the formal properties of our fuzzy Rand index, we show that it exhibits desirable metrical properties.
Fuzzy Modeling of Labeled Point Cloud Superposition for the Comparison of Protein Binding Sites.
Abstract — Geometric objects are often represented approximately in terms of a finite set of points in three-dimensional Euclidean space. In this paper, we extend this representation to what we call labeled point clouds. A labeled point cloud is a finite set of points, where each point is not only associated with a position in three-dimensional space, but also with a discrete class label that represents a specific property. This type of model is especially suitable for modeling biomolecules such as proteins and protein binding sites, where a label may represent an atom type or a physicochemical property. Proceeding from this representation, we address the question of how to compare two labeled points clouds in terms of similarity. Using fuzzy modeling techniques, we develop a suitable similarity measure as well as an efficient evolutionary algorithm to compute it. Finally, an application study is presented in which the approach is used to classify protein binding sites.
Fuzzy Modeling of Labeled Point Cloud Superposition for the Comparison of Protein Binding Sites.
Abstract — Geometric objects are often represented approximately in terms of a finite set of points in three-dimensional Euclidean space. In this paper, we extend this representation to what we call labeled point clouds. A labeled point cloud is a finite set of points, where each point is not only associated with a position in three-dimensional space, but also with a discrete class label that represents a specific property. This type of model is especially suitable for modeling biomolecules such as proteins and protein binding sites, where a label may represent an atom type or a physicochemical property. Proceeding from this representation, we address the question of how to compare two labeled points clouds in terms of similarity. Using fuzzy modeling techniques, we develop a suitable similarity measure as well as an efficient evolutionary algorithm to compute it. Finally, an application study is presented in which the approach is used to classify protein binding sites.
Similarity Analysis of Protein Binding Sites A Generalization of the Maximum Common Subgraph Measure Based on Quasi%2DClique Detection.

A New Instance%2DBased Label Ranking Approach Using the Mallows Model.

Efficient Construction of Multiple Geometrical Alignments for the Comparison of Protein Binding Sites.

A New Instance%2DBased Label Ranking Approach Using the Mallows Model.



FURIA an algorithm for unordered fuzzy rule induction.

Prelude to the papers %22Fuzzy case based reasoning for facial expression recognition%22 and %22Temporal similarity by measuring possibilistic uncertainty in CBR%22.

FR3 A Fuzzy Rule Learner for Inducing Reliable Classifiers.

Fuzzy Operator Trees for Modeling Rating Functions.

Supporting Case%2DBased Retrieval by Similarity Skyline.

Combining instance%2Dbased learning and logistic regression for multilabel classification.
Abstract. Multilabel classification is an extension of conventional classification in which a single instance can be associated with multiple labels. Recent research has shown that, just like for standard classification, instance-based learning algorithms relying on the nearest neighbor estimation principle can be used quite successfully in this context. However, since hitherto existing algorithms do not take correlations and interdependencies between labels into account, their potential has not yet been fully exploited. In this paper, we propose a new approach to multilabel classification, which is based on a framework that unifies instancebased learning and logistic regression, comprising both methods as special cases. This approach allows one to capture interdependencies between labels and, moreover, to combine model-based and similarity-based inference for multilabel classification. As will be shown by experimental studies, our approach is able to improve predictive accuracy in terms of several evaluation criteria for multilabel prediction. 1
Learning Similarity Functions from Qualitative Feedback.

Supporting Case%2DBased Retrieval by Similarity Skylines Basic Concepts and Extensions.

Preface Uncertainty%2C Similarity%2C and Knowledge Discovery in CBR.

Instance%2DBased Label Ranking using the Mallows Model.

A Critical Analysis of Variants of the AUC.

Evolutionary Construction of Multiple Graph Alignments for the Structural Analysis of Biomolecules.

Evolutionary Construction of Multiple Graph Alignments for Mining Structured Biomolecular Data.

Weighted Voting as Approximate MAP Prediction in Pairwise Classification.


Fuzzy Methods for Data Mining and Machine Learning State ofthe Art and Prospects.


Learning valued preference structures for solving classification problems.

Is an ordinal class structure useful in classifier learning%3F

Case%2Dbased learning in a bipolar possibilistic framework.

A critical analysis of variants of the AUC.

Multilabel classification via calibrated label ranking.
Abstract Label ranking studies the problem of learning a mapping from instances to rankings over a predefined set of labels. Hitherto existing approaches to label ranking implicitly operate on an underlying (utility) scale which is not calibrated in the sense that it lacks a natural zero point. We propose a suitable extension of label ranking that incorporates the calibrated scenario and substantially extends the expressive power of these approaches. In particular, our extension suggests a conceptually novel technique for extending the common learning by pairwise comparison approach to the multilabel scenario, a setting previously not being amenable to the pairwise decomposition technique. The key idea of the approach is to introduce an artificial calibration label that, in each example, separates the relevant from the irrelevant labels. We show that this technique can be viewed as a combination of pairwise preference learning and the conventional relevance classification technique, where a separate classifier is trained to predict whether a label is relevant or not. Empirical results in the area of text categorization, image classification and gene analysis underscore the merits of the calibrated model in comparison to state-of-the-art multilabel learning methods.
Case%2DBased Approximate Reasoning

On Pairwise Naive Bayes Classifiers.

On Minimizing the Position Error in Label Ranking.

Fuzzy%2DRelational Classification Combining Pairwise Decomposition Techniques with Fuzzy Preference Modeling.
This paper introduces a new approach to classification which combines pairwise decomposition techniques from machine learning with ideas and tools from fuzzy preference modeling. The approach, called fuzzy relational classification, effectively reduces the problem of classification to a problem of decision making based on a fuzzy preference relation. It will be shown that, by decomposing sucharelationintoastrict preference, an indifference, and an incomparability relation, it becomes possible to quantify different types of uncertainty in classification, and thereby to support more sophisticated classification and postprocessing strategies.
Adaptive Optimization of the Number of Clusters in Fuzzy Clustering.


Graph Alignment Fuzzy Pattern Mining for the Structural Analysis of Protein Active Sites.






Interactive Ranking of Skylines Using Machine Learning Techniques.
We thank our sponsors:
Risk assessment system of natural hazards A new approach based on fuzzy probability.

Credible Case%2DBased Inference Using Similarity Profiles.

In Defense of Fuzzy Association Analysis.

Multiple Graph Alignment for the Structural Analysis of Protein Active Sites.

Comparing probability measures using possibility theory A notion of relative peakedness.

Efficient instance%2Dbased learning on data streams.

Editorial.
calls &amp; calendarEDITORIAL
A fuzzy%2Dprobabilistic earthquake risk assessment system.

Special Issue on Soft Computing for Information Mining.

A Unified Model for Multilabel Classification and Ranking.
 Label ranking studies the problem of learning a mapping from instances to rankings over a predefined set of labels. Hitherto existing approaches to label ranking implicitly operate on an underlying (utility) scale which is not calibrated in the sense that it lacks a natural zero point. We propose a suitable extension of label ranking that incorporates the calibrated scenario and substantially extends the expressive power of these approaches. In particular, our extension suggests a conceptually novel technique for extending the common learning by pairwise comparison approach to the multilabel scenario, a setting previously not being amenable to the pairwise decomposition technique. We present empirical results in the area of text categorization and gene analysis, underscoring the merits of the calibrated model in comparison to state-of-the-art multilabel learning methods. 
Case%2DBased Label Ranking.
The problem of label ranking has recently been introduced as an extension of conventional classification in the field of machine learning. In this paper, we argue that label ranking is an amenable task from a CBR point of view and, in particular, is more amenable to supporting case-based problem solving than standard classification. Moreover, by developing a case-based approach to label ranking, we will show that, the other way round, concepts and techniques from CBR are also useful for label ranking. In addition to an experimental study in which case-based label ranking is compared to conventional nearest neighbor classification, we present an application in which label ranking is used for node ordering in heuristic search. 
Hierarchical Classification by Expected Utility Maximization.

Online clustering of parallel data streams.
In recent years, the management and processing of so-called data streams has become a topic of active research in several fields of computer science such as, e.g., distributed systems, database systems, and data mining. A data stream can roughly be thought of as a transient, continuously increasing sequence of time-stamped data. In this paper, we consider the problem of clustering parallel streams of real-valued data, that is to say, continuously evolving time series. In other words, we are interested in grouping data streams the evolution over time of which is similar in a specific sense. In order to maintain an up-to-date clustering structure, it is necessary to analyze the incoming data in an online manner, tolerating not more than a constant time delay. For this purpose, we develop an efficient online version of the classical K-means clustering algorithm. Our method’s efficiency is mainly due to a scalable online transformation of the original data which allows for a fast computation of approximate distances between streams. Key words: data mining, clustering, data streams, fuzzy sets 1
A systematic approach to the assessment of fuzzy association rules.
In order to allow for the analysis of data sets including numerical attributes, several generalizations of association rule mining based on fuzzy sets have been proposed in the literature. While the formal specification of fuzzy associations is more or less straightforward, the assessment of such rules by means of appropriate quality measures is less obvious. Particularly, it assumes an understanding of the semantic meaning of a fuzzy rule. This aspect has been ignored by most existing proposals, which must therefore be considered as ad-hoc to some extent. In this paper, we develop a systematic approach to the assessment of fuzzy association rules. To this end, we proceed from the idea of partitioning the data stored in a database into examples of a given
Learning from ambiguously labeled examples.
Abstract. Inducing a classification function from a set of examples in the form of labeled instances is a standard problem in supervised machine learning. In this paper, we are concerned with ambiguous label classification (ALC), an extension of this setting in which several candidate labels may be assigned to a single example. By extending three concrete classification methods to the ALC setting and evaluating their performance on benchmark data sets, we show that appropriately designed learning algorithms can successfully exploit the information contained in ambiguously labeled examples. Our results indicate that the fundamental idea of the extended methods, namely to disambiguate the label information by means of the inductive bias underlying (heuristic) machine learning methods, works well in practice. 1
Fuzzy methods for case%2Dbased recommendation and decision support.

A Notion of Comparative Probabilistic Entropy Based on the Possibilistic Specificity Ordering.

Improving the interpretability of data%2Ddriven evolving fuzzy systems.
This paper develops methods for reducing the complexity and, thereby, improving the linguistic interpretability of Takagi-Sugeno fuzzy systems that are learned online in a data-driven, incremental way. In order to ensure the transparency of the evolving fuzzy system at any time, complexity reduction must be performed in an online mode as well. Our methods are evaluated on high-dimensional data coming from an industrial measuring process.
Cho%2Dk%2DNN A Method for Combining Interacting Pieces of Evidence in Case%2DBased Learning.
The case-based learning paradigm relies upon memorizing cases in the form of successful problem solving experience, such as e.g. a pattern along with its classification in pattern recognition or a problem along with a solution in case-based reasoning. When it comes to solving a new problem, each of these cases serves as an individual piece of evidence that gives an indication of the solution to that problem. In this paper, we elaborate on issues concerning the proper combination (aggregation) of such pieces of evidence. Particularly, we argue that cases retrieved from a case library must not be considered as independent information sources, as implicitly done by most case-based learning methods. Focusing on the problem of prediction as a performance task, we propose a new inference principle that combines potentially interacting pieces of evidence by means of the so-called (discrete) Choquet-integral. Our method, called Cho-k-NN, takes interdependencies between the stored cases into account and can be seen as a generalization of weighted nearest neighbor estimation. 1
Learning Complexity%2DBounded Rule%2DBased Classifiers by Combining. Association Analysis and Genetic Algorithms.
We propose a method for learning rule-based classifiers that can be seen as a compromise between a complete enumeration of the hypothesis space and heuristic strategies to search this space in a greedy manner. The method consists of two main parts: In a first step, a sufficiently large number of individual, high-quality classification rules is generated. In a second step, a classifier is assembled from the candidate rules thus obtained. This comes down to selecting a proper subset of these rules, a combinatorial problem that we shall approach by means of genetic algorithms. For the candidate generation step, we suggest using association rule mining. Apart from learning accurate classifiers, a main motivation of our method is the possibility to control the tradeoff between accuracy and transparency (complexity) of a model in a more explicit way.
Learning from Ambiguously Labeled Examples.
Abstract. Inducing a classification function from a set of examples in the form of labeled instances is a standard problem in supervised machine learning. In this paper, we are concerned with ambiguous label classification (ALC), an extension of this setting in which several candidate labels may be assigned to a single example. By extending three concrete classification methods to the ALC setting and evaluating their performance on benchmark data sets, we show that appropriately designed learning algorithms can successfully exploit the information contained in ambiguously labeled examples. Our results indicate that the fundamental idea of the extended methods, namely to disambiguate the label information by means of the inductive bias underlying (heuristic) machine learning methods, works well in practice. 1
Learning Label Preferences Ranking Error Versus Position Error.
We consider the problem of learning a ranking function,that is a mapping from instances to rankings over a finite number of labels. Our learning method, referred to as ranking by pairwise comparison (RPC), first induces pairwise order relations from suitable training data, using a natural extension of so-called pairwise classification. A ranking is then derived from a set of such relations by means of a ranking procedure. This paper elaborates on a key advantage of such a decomposition, namely the fact that our learner can be adapted to different loss functions by using different ranking procedures on the same underlying order relations. In particular, the Spearman rank correlation is minimized by using a simple weighted voting procedure. Moreover, we discuss a loss function suitable for settings where candidate labels must be tested successively until a target label is found, and propose a ranking procedure for minimizing the corresponding risk.  
Cho%2Dk%2DNN A Method for Combining Interacting Pieces of Evidence in Case%2DBased Learning.
The case-based learning paradigm relies upon memorizing cases in the form of successful problem solving experience, such as e.g. a pattern along with its classification in pattern recognition or a problem along with a solution in case-based reasoning. When it comes to solving a new problem, each of these cases serves as an individual piece of evidence that gives an indication of the solution to that problem. In this paper, we elaborate on issues concerning the proper combination (aggregation) of such pieces of evidence. Particularly, we argue that cases retrieved from a case library must not be considered as independent information sources, as implicitly done by most case-based learning methods. Focusing on the problem of prediction as a performance task, we propose a new inference principle that combines potentially interacting pieces of evidence by means of the so-called (discrete) Choquet-integral. Our method, called Cho-k-NN, takes interdependencies between the stored cases into account and can be seen as a generalization of weighted nearest neighbor estimation. 1
On Position Error and Label Ranking through Iterated Choice.
We consider the problem of learning a ranking  function, that is a mapping from instances to  rankings over a finite number of labels. Our  learning method, referred to as ranking by pairwise  comparison (RPC), first induces pairwise  order relations from suitable training data, using  a natural extension of so-called pairwise classification. A ranking is
Clustering of gene expression data using a local shape%2Dbased similarity measure.

Special issue on fuzzy sets in knowledge discovery.

Fuzzy methods in machine learning and data mining Status and prospects.

Experience%2DBased Decision Making A Satisficing Decision Tree Approach.

Preference Learning.
In this paper, we propose a probabilistic kernel approach to preference learning based on Gaussian processes. A new likelihood function is proposed to capture the preference relations in the Bayesian framework. The generalized formulation is also applicable to tackle many multiclass problems. The overall approach has the advantages of Bayesian methods for model selection and probabilistic prediction. Experimental results compared against the constraint classification approach on several benchmark datasets verify the usefulness of this algorithm. 1.
Instance%2DBased Prediction with Guaranteed Confidence.
Instance-based learning (IBL) algorithms have proved to be successful in many applications. However, as opposed to standard statistical methods, a prediction in IBL is usually given without characterizing its confidence. In this paper, we propose an IBL method that allows for deriving set-valued predictions that cover the correct answer (label) with high probability. Our method makes use of a formal model of the heuristic inference principle suggesting that similar instances do have similar labels. The focus of this paper is on the prediction of numeric values (regression), even though the method is also useful for classification problems if a reasonable similarity measure can be defined on the set of classes.
Fuzzy Methods in Knowledge Discovery.

Graph Alignments A New Concept to Detect Conserved Regions in Protein Active Sites.

Efficient similarity search in protein structure databases by k%2Dclique hashing.

Flexible constraints for regularization in learning from data.


26. Jahrestagung K%FCnstliche Intelligenz %28KI%2D2003%29.

Pairwise Preference Learning and Ranking.
We consider supervised learning of a ranking function, which is a mapping  from instances to total orders over a set of labels (options). The training  information consists of examples with partial (and possibly inconsistent) information  about their associated rankings. From these, we induce a ranking  function by reducing the original problem to a number of binary classification  problems, one for each pair of labels. The main objective of this work is to  investigate the trade-off between the quality of the induced ranking function  and the computational complexity of the algorithm, both depending on the  amount of preference information given for each example. To this end, we  present theoretical results on the complexity of pairwise preference learning.
Instance%2Dbased collaborative filtering with fuzzy labels.
In recommender systems, user ratings of items are often represented in terms of linguistic labels such as “fair ” or “very good”. We investigate the potential of fuzzy sets as a means for modeling such labels, especially with regard to collaborative filtering algorithms. We propose a related fuzzy version of instance-based (memory-based) collaborative filtering and argue that it leads to more informative and accurate predictions. These claims are validated by means of two experimental studies.

Regularized Learning with Flexible Constraints.

A Note on Quality Measures for Fuzzy Asscociation Rules.

Inducing Fuzzy Concepts through Extended Version Space Learning.

On the representation of fuzzy rules in terms of crisp rules.

Instance%2DBased Learning of Credible Label Sets.

Possibilistic instance%2Dbased learning.

On the representation of fuzzy rules in terms of crisp rules.

Association Rules for Expressing Gradual Dependencies.

Flexibility and Fuzzy Case%2DBased Evaluation in Querying An Illustration in an Experimental Setting.

On the Representation and Combination of Evidence in Instance%2DBased Learning.

Possibilistic Induction in Decision%2DTree Learning.

A Fuzzy Approach to Flexible Case%2Dbased Querying Methodology and Experimentation.

Association Rules for Expressing Gradual Dependencies.

Fuzzy Association Rules Semantic Issues and Quality Measures.

Implication%2DBased Fuzzy Association Rules.
Abstract — Fuzzy association rules described by the natural language are well suited for the thinking of human subject and will help to increase the flexibility for supporting user in maki ng de c i si ons or de si gni ng the fuzzy sy ste ms. However, the efficiency of algorithms needs to be improved to handle huge datasets in real word. in this paper, we present an efficient algorithm named Fuzzy Cluster-Based Association Rules(FCBAR). The FCBAR method is to create cluster tables by scanning the database once, and then clustering the transaction records to the k_th cluster table, where the length of a record is k. Moreover, the fuzzy large itemsets are generated by contrasts with the partial cluster tables. This prunes considerable amount of data, reduces the time needed to perform data scans and requires less contrast. Experiments with the real-life database show that FCBAR outperforms fuzzy Apriori_like algorithm, a well–known and widely used association rules algorithm. Index Terms —Cluster Table, Fuzzy Association Rules 1.
Similarity%2Dbased inference as evidential reasoning.

Change Detection in Heuristic Search.
The order in which nodes are explored in a (depth-first) iterative deepening search strategy is principally determined by the condition under which a path of the search tree is cut off in each search phase. A corresponding criterion, which has a strong influence on the performance of the overall (heuristic) search procedure, is generally realized in the form of an upper cost bound. In this paper, we develop an effective and computationally efficient termination criterion based on statistical methods of change detection. The criterion is local in the sense that it depends on properties of a path itself, rather than on the comparison with other paths. Loosely speaking, the idea is to take a systematic change in the (heuristic) evaluation of nodes along a search path as an indication of suboptimality. An expected utility criterion which also takes the consequence of the suboptimal search decision on the solution quality into account is proposed as a generalization of this idea.
Focusing Search by Using Problem Solving Experience.

Similarity%2Dbased Inference as Evitential Reasoning.

A Method for Predicting Solutions in Case%2DBased Problem Solving.

Flexible Control of Case%2DBased Prediction in the Framework of Possibility Theory.

A Possibilistic Formalization of Case%2DBased Reasoning and Decision Making.

Exploiting Similarity for Supporting Data Analysis and Problem Solving.

Toward a Probabilistic Formalization of Case%2DBased Inference.

Numerical Methods for Fuzzy Initial Value Problems.
this paper, we propose an approach to modelling and simulation of uncertain dynamics which is based on the following ideas: We consider (linguistic) descriptions of uncertain functional relationships characterizing the behavior of some dynamical system. Based on a certain interpretation of such rule-based models, we derive a fuzzy function
An Approach to Modelling and Simulation of Uncertain Dynamical Systems.
this paper, we propose an approach to modelling and simulation of uncertain dynamics which is based on the following ideas: We consider (linguistic) descriptions of uncertain functional relationships characterizing the behavior of some dynamical system. Based on a certain interpretation of such rule-based models, we derive a fuzzy function
