Toward the frontiers of stacked generalization architecture for learning

Download

index.pdf

Date

2007

Author

Mertayak, Cüneyt

Metadata

Show full item record

Item Usage Stats

147
views

43
downloads

In pattern recognition, “bias-variance” trade-off is a challenging issue that the scientists has been working to get better generalization performances over the last decades. Among many learning methods, two-layered homogeneous stacked generalization has been reported to be successful in the literature, in different problem domains such as object recognition and image annotation. The aim of this work is two-folded. First, the problems of stacked generalization are attacked by a proposed novel architecture. Then, a set of success criteria for stacked generalization is studied. A serious drawback of stacked generalization architecture is the sensitivity to curse of dimensionality problem. In order to solve this problem, a new architecture named “unanimous decision” is designed. The performance of this architecture is shown to be comparably similar to two layered homogeneous stacked generalization architecture in low number of classes while it performs better than stacked generalization architecture in higher number of classes. Additionally, a new success criterion for two layered homogeneous stacked generalization architecture is proposed based on the individual properties of the used descriptors and it is verified in synthetic datasets.

Subject Keywords

Computer Software.

URI

http://etd.lib.metu.edu.tr/upload/3/12608793/index.pdf
https://hdl.handle.net/11511/17079

Collections

Graduate School of Natural and Applied Sciences, Thesis

Suggestions

OpenMETU
Core

A systematic study of probabilistic aggregation strategies in swarm robotic systems Soysal, Onur; Şahin, Erol; Department of Computer Engineering (2005) In this study, a systematic analysis of probabilistic aggregation strategies in swarm robotic systems is presented. A generic aggregation behavior is proposed as a combination of four basic behaviors: obstacle avoidance, approach, repel, and wait. The latter three basic behaviors are combined using a three-state finite state machine with two probabilistic transitions among them. Two different metrics were used to compare performance of strategies. Through systematic experiments, how the aggregation performa...
Modelling and predicting binding affinity of PCP-like compounds using machine learning methods Erdaş, Özlem; Alpaslan, Ferda Nur; Department of Computer Engineering (2007) Machine learning methods have been promising tools in science and engineering fields. The use of these methods in chemistry and drug design has advanced after 1990s. In this study, molecular electrostatic potential (MEP) surfaces of PCP-like compounds are modelled and visualized in order to extract features which will be used in predicting binding affinity. In modelling, Cartesian coordinates of MEP surface points are mapped onto a spherical self-organizing map. Resulting maps are visualized by using values...
An automated conversion of temporal databases into XML with fuzziness option Işıkman, Ömer Özgün; Polat, Faruk; Özyer, Tansel; Department of Computer Engineering (2010) The importance of incorporating time in databases has been well realized by the community and time varying databases have been extensively studied by researchers. The main idea is to model up-to-date changes to data since it became available. Time information is mostly overlaid on the traditional databases, and extensional time dimension helps in inquiring or past data; this all becomes possible only once the idea is realized and favored by commercial database management systems. Unfortunately, one disadvan...
Memetic algorithms for timerabling problems in pricate schools Aldoğan, Deniz; Alpaslan, Ferda Nur; Department of Computer Engineering (2005) The aim of this study is to introduce a real-world timetabling problem that exists in some private schools in Turkey and to solve such problem instances utilizing memetic algorithms. Being a new type of problem and for privacy reasons, there is no real data available. Hence for benchmarking purposes, a random data generator has been implemented. Memetic algorithms (MAs) combining genetic algorithms and hill-climbing are applied to solve synthetic problem instances produced by this generator. Different types...
Ontology population using human computation Evirgen, Gencay Kemal; Alpaslan, Ferda Nur; Department of Computer Engineering (2010) In recent years, many researchers have developed new techniques on ontology population. However, these methods cannot overcome the semantic gap between humans and the extracted ontologies. Words-Around is a web application that forms a user-friendly environment which channels the vast Internet population to provide data towards solving ontology population problem that no known efficient computer algorithms can yet solve. This application’s fundamental data structure is a list of words that people naturally ...

Citation Formats

C. Mertayak, “Toward the frontiers of stacked generalization architecture for learning,” M.S. - Master of Science, Middle East Technical University, 2007.