The package JavaEvA (a Java implementation of Evolutionary Algorithms) is a general modular framework with an inherent client server structure to solve practical optimization problems. This package was especially designed to test and develop new approaches for Evolutionary Algorithms and to utilize them in real-world applications. JavaEvA already provides implementations of the most common Evolutionary Algorithms, like Genetic Algorithms, CHC Adaptive Search, Population Based Incremental Learning, Evolution Strategies, Model-Assisted Evolution Strategies, Genetic Programming and Grammatical Evolution. In addition the modular framework of JavaEvA allows everyone to add their own optimization modules to meet their specific requirements. The JavaEvA package uses a generic GUI framework that allows GUI access to any member of a class if get and set methods are provided and an editor is defined for the given data type. This approach allows very fast development cycles, since hardly any additional effort is necessary for implementing GUI elements, while still at the same time user specific GUI elements can be developed and integrated to increase usability. Since we cannot anticipate specific optimization problem and requirements, it is necessary for users to define their optimization problem. Therefore, we provide an additional framework and explain how one can include JavaEvA in an existing Java project or how one can implement ones own optimization problem and optimize it by using JavaEvA. This gives users total control of the optimization algorithms used.
[pdf]
@techreport{2005_70,
author = {Streichert, Felix and Ulmer, Holger},
title = {{JavaEvA}: a {Java} based framework for Evolutionary Algorithms},
institution = {Center for Bioinformatics Tuebingen (ZBIT) of the Eberhard Karls
University, Tuebingen, Germany},
year = {2005},
number = {WSI-2005-06},
month = apr,
abstract = {The package JavaEvA (a Java implementation of Evolutionary Algorithms)
is a general modular framework with an inherent client server structure
to solve practical optimization problems. This package was especially
designed to test and develop new approaches for Evolutionary Algorithms
and to utilize them in real-world applications. JavaEvA already provides
implementations of the most common Evolutionary Algorithms, like
Genetic Algorithms, CHC Adaptive Search, Population Based Incremental
Learning, Evolution Strategies, Model-Assisted Evolution Strategies,
Genetic Programming and Grammatical Evolution. In addition the modular
framework of JavaEvA allows everyone to add their own optimization
modules to meet their specific requirements. The JavaEvA package
uses a generic GUI framework that allows GUI access to any member
of a class if get and set methods are provided and an editor is defined
for the given data type. This approach allows very fast development
cycles, since hardly any additional effort is necessary for implementing
GUI elements, while still at the same time user specific GUI elements
can be developed and integrated to increase usability. Since we cannot
anticipate specific optimization problem and requirements, it is
necessary for users to define their optimization problem. Therefore,
we provide an additional framework and explain how one can include
JavaEvA in an existing Java project or how one can implement ones
own optimization problem and optimize it by using JavaEvA. This gives
users total control of the optimization algorithms used.},
url = {http://w210.ub.uni-tuebingen.de/dbt/volltexte/2005/1702/}
}
@article{2005_7,
author = {Wegner, J\"org K. and Sieker, Florian and Zell, Andreas},
title = {Relevance of feature selection for clustering molecules},
journal = {Abstracts of Papers American Chemical Society (229th National Meeting
of the American-Chemical-Society)},
year = {2005},
volume = {229},
pages = {U612},
number = {Part 1},
month = mar,
abstract = {We present an extensive study to classify and cluster four different
activity classes (5HT1A antagonists, thrombin inhibitors, MAO inhibitors
and H2 antagonists) by using four different feature selection algorithms,
ten different classification algorithms and three clustering algorithms.
We show that depending on the used number of features the generalization
ability of the results varies dramatically for clustering compounds.
The number of features used was based on the previous feature selection
results. The classification rates ranges from 85% up to 97%, which
are much better than the clustering confusion matrix results. Finally,
we conclude the work with presenting a new approch avoiding the feature
selection dilemma.},
address = {San Diego, CA, USA},
}