V. Popova (Viara)
http://repub.eur.nl/ppl/390/
List of Publicationsenhttp://repub.eur.nl/eur_signature.png
http://repub.eur.nl/
RePub, Erasmus University RepositoryMonotone classification by function decomposition
http://repub.eur.nl/pub/52796/
Thu, 01 Dec 2005 00:00:01 GMT<div>V. Popova</div><div>J.C. Bioch</div>
The paper focuses on the problem of classification by function decomposition within the frame of monotone classification, We propose a decomposition method for discrete functions which can be applied to monotone problems in order to generate a monotone classifier based on the extracted concept hierarchy. We formulate and prove a criterion for the existence of a positive extension of the scheme f = g(So,h(S1)) in the context of discrete functions. We also propose a method for finding an assignment for the intermediate concept with a minimal number of values.Knowledge Discovery and Monotonicity
http://repub.eur.nl/pub/1201/
Thu, 01 Apr 2004 00:00:01 GMT<div>V. Popova</div>
The monotonicity property is ubiquitous in our lives and it appears in different roles: as domain knowledge, as a requirement, as a property that reduces the complexity of the problem, and so on. It is present in various domains: economics, mathematics, languages, operations research and many others. This thesis is focused on the monotonicity property in knowledge discovery and more specifically in classification, attribute reduction, function decomposition, frequent patterns generation and missing values handling. Four specific problems are addressed within four different methodologies, namely, rough sets theory, monotone decision trees, function decomposition and frequent patterns generation. In the first three parts, the monotonicity is domain knowledge and a requirement for the outcome of the classification process. The three methodologies are extended for dealing with monotone data in order to be able to guarantee that the outcome will also satisfy the monotonicity requirement. In the last part, monotonicity is a property that helps reduce the computation of the process of frequent patterns generation. Here the focus is on two of the best algorithms and their comparison both theoretically and experimentally.
About the Author:
Viara Popova was born in Bourgas, Bulgaria in 1972. She followed her secondary
education at Mathematics High School "Nikola Obreshkov" in Bourgas. In 1996
she finished her higher education at Sofia University, Faculty of Mathematics
and Informatics where she graduated with major in Informatics and specialization
in Information Technologies in Education. She then joined the Department
of Information Technologies,
First as an associated member and from 1997 as an assistant professor.
In 1999 she became a PhD student at Erasmus University Rotterdam, Faculty
of Economics, Department of Computer Science. In 2004 she joined the
Artificial Intelligence Group within the Department of Computer Science, Faculty
of Sciences at Vrije Universiteit Amsterdam as a PostDoc researcher.This thesis is positioned in the area of knowledge discovery with special attention to problems where the property of monotonicity plays an important role. Monotonicity is a ubiquitous property in all areas of life and has therefore been widely studied in mathematics. Monotonicity in knowledge discovery can be treated as available background information that can facilitate and guide the knowledge extraction process. While in some sub-areas methods have already been developed for taking this additional information into account, in most methodologies it has not been extensively studied or even has not been addressed at all. This thesis is a contribution to a change in that direction. In the thesis, four specific problems have been examined from different sub-areas of knowledge discovery: the rough sets methodology, monotone decision trees, function decomposition and frequent patterns discovery. In the first three parts, the monotonicity is domain knowledge and a requirement for the outcome of the classification process. The three methodologies are extended for dealing with monotone data in order to be able to guarantee that the outcome will also satisfy the monotonicity requirement. In the last part, monotonicity is a property that helps reduce the computation of the process of frequent patterns generation. Here the focus is on two of the best algorithms and their comparison both theoretically and experimentally.Induction of Ordinal Decision Trees
http://repub.eur.nl/pub/271/
Mon, 10 Feb 2003 00:00:01 GMT<div>J.C. Bioch</div><div>V. Popova</div>
This paper focuses on the problem of monotone decision trees from the
point of view of the multicriteria decision aid methodology (MCDA). By
taking into account the preferences of the decision maker, an attempt is
made to bring closer similar research within machine learning and MCDA.
The paper addresses the question how to label the leaves of a tree
in a way that guarantees the monotonicity of the resulting tree. Two
approaches are proposed for that purpose - dynamic and static labeling
which are also compared experimentally.
The paper further considers the problem of splitting criteria in the con-
text of monotone decision trees. Two criteria from the literature are com-
pared experimentally - the entropy criterion and the number of con
criterion - in an attempt to find out which one fits better the specifics of
the monotone problems and which one better handles monotonicity noise.Monotone Decision Trees and Noisy Data
http://repub.eur.nl/pub/207/
Mon, 17 Jun 2002 00:00:01 GMT<div>J.C. Bioch</div><div>V. Popova</div>
The decision tree algorithm for monotone classification presented in [4, 10] requires strictly monotone data sets. This paper addresses the problem of noise due to violation of the monotonicity constraints and proposes a modification of the algorithm to handle noisy data. It also presents methods for controlling the size of the resulting trees while keeping the monotonicity property whether the data set is monotone or not.Bankruptcy Prediction with Rough Sets
http://repub.eur.nl/pub/76/
Thu, 22 Feb 2001 00:00:01 GMT<div>J.C. Bioch</div><div>V. Popova</div>
The bankruptcy prediction problem can be considered an or
dinal classification problem. The classical theory of Rough Sets describes
objects by discrete attributes, and does not take into account the order-
ing of the attributes values. This paper proposes a modification of the
Rough Set approach applicable to monotone datasets. We introduce re-
spectively the concepts of monotone discernibility matrix and monotone
(object) reduct. Furthermore, we use the theory of monotone discrete
functions developed earlier by the first author to represent and to com-
pute decision rules. In particular we use monotone extensions, decision
lists and dualization to compute classification rules that cover the whole
input space. The theory is applied to the bankruptcy prediction problem.