[Desire-commits] r7 - papers/useR_2008

Tue Apr 29 10:17:42 CEST 2008

Author: olafm
Date: 2008-04-29 10:17:38 +0200 (Tue, 29 Apr 2008)
New Revision: 7

Modified:
   papers/useR_2008/useR_2008-abstract.bib
   papers/useR_2008/useR_2008-abstract.pdf
   papers/useR_2008/useR_2008-abstract.tex
Log:
* New abstract with focus on optimization

Modified: papers/useR_2008/useR_2008-abstract.bib
===================================================================

--- papers/useR_2008/useR_2008-abstract.bib	2008-03-31 09:06:48 UTC (rev 6)
+++ papers/useR_2008/useR_2008-abstract.bib	2008-04-29 08:17:38 UTC (rev 7)
@@ -28,7 +28,7 @@
 }
  
 @phdthesis{steuer2005,
-author = {Detlef Steuer},
+author = {D.~Steuer},
 title = {Statistische Eigenschaften der Multikriteriellen Optimierung mittels Wünschbarkeiten},
 school = {Technische Universit{\"a}t Dortmund},
 year = 2005,
@@ -36,7 +36,7 @@
 }
  
 @phdthesis{trautmann2004,
-author={H. Trautmann},
+author={H.~Trautmann},
 title={{Q}ualit{\"a}tskontrolle in der {I}ndustrie anhand von {K}ontrollkarten f{\"u}r
        {W}{\"u}nschbarkeitsindizes -- {A}nwendungsfeld {L}agerverwaltung},
 school={Technische Universit{\"a}t Dortmund},
@@ -45,7 +45,7 @@
 }
 
 @techreport{trautmann2004a,
-author={H. Trautmann and C. Weihs},
+author={H.~Trautmann and C.~Weihs},
 title={{U}ncertainty of the {O}ptimum {I}nfluence {F}actor {L}evels in {M}ulticriteria
         {O}ptimization {U}sing the {C}oncept of {D}esirability},
 institution={{SFB 475}, {S}tatistics {F}aculty, Technische Universität Dortmund, {G}ermany},
@@ -54,7 +54,7 @@
 }
 
 @article{trautmann2006,
-author={Heike Trautmann and Claus Weihs},
+author={H.~Trautmann and C.~Weihs},
 title={On the distribution of the desirability index using {H}arrington's desirability function},
 journal={Metrika},
 year=2006,
@@ -62,3 +62,14 @@
 number={2},
 pages={207--213}
 }
+
+Deb, K., Pratap, A., and Agarwal, S.. A Fast and Elitist Multiobjective Genetic Algorithm: NSGA-II. IEEE Transactions on Evolutionary Computation, 6 (8) (2002), 182-197.
+ at article{dep2002,
+author = {K. Deb and A. Pratap and S. Agarwal},
+title = {A {F}ast and {E}litist {M}ultiobjectiv {G}enetic {A}lgorithm: {NSGA-II}},
+journal={IEEE Transactions on Evolutionary Computation},
+year=2002,
+volume=6,
+number=8,
+pages={182--197}
+}

Modified: papers/useR_2008/useR_2008-abstract.pdf
===================================================================
(Binary files differ)

Modified: papers/useR_2008/useR_2008-abstract.tex
===================================================================
--- papers/useR_2008/useR_2008-abstract.tex	2008-03-31 09:06:48 UTC (rev 6)
+++ papers/useR_2008/useR_2008-abstract.tex	2008-04-29 08:17:38 UTC (rev 7)
@@ -7,7 +7,7 @@
  
 \hyphenation{Tech-ni-sche}
 \begin{document}
-\title{\LARGE How to choose the appropriate Object-Oriented framework \\
+\title{\LARGE Desirabilitiy functions in multicriteria optimization \\
   \LARGE Observations made while implementing \texttt{desiRe}}
 \author{%
   \large Olaf Mersmann$^{\text{1}}$
@@ -22,35 +22,38 @@
 \footnotetext[1]{Fakultät Statistik, Technische Universität Dortmund}
 \footnotetext[2]{Fakultät WiSo, Helmut-Schmidt Universität Hamburg}
  
-\vskip-3em The talk will focus on some of the challenges faced during the development of
-the \texttt{desiRe} package. Desirability functions and desirability indices are powerful
-tools for multicriteria optimization und multicriteria quality control purposes. The
-package desiRe not only provides functions for computing desirability functions of
-Harrington- \citep{harrington1965} and Derringer/Suich-type \citep{derringer1980} but also
-allows the specification of the functions in an interactive way. Density and distribution
-functions of the desirability functions and the desirabi\-lity index are integrated
-including the possibility of random number generation \citep{steuer2005},
-\citep{trautmann2006}. Optimization procedures for the desirability index and a method for
-determining the uncertainty of the optimum influence factor levels \citep{trautmann2004a}
-as wells as a control chart for the desirability index with analysis of out-of
-control-signals are implemented \citep{trautmann2004}. The Desirability Pareto-Concept
-allows focussing on relevant parts of the Pareto-front by integrating
-a-priori-expert-knowledge in the multicriteria optimization process \citep{mehnen2007}.
+\vskip-3em Desirability functions and desirability indices are
+powerful tools for multicriteria optimization und multicriteria
+quality control purposes. The package \texttt{desiRe} not only
+provides functions for computing desirability functions of Harrington-
+\citep{harrington1965} and Derringer/Suich-type \citep{derringer1980}
+but also allows the specification of functions in an interactive
+manner. Density and distribution functions of the desirability
+functions and the desirabi\-lity index are integrated including the
+possibility of random number generation \citep{steuer2005},
+\citep{trautmann2006}. Optimization procedures for the desirability
+index and a method for determining the uncertainty of the optimum
+influence factor levels \citep{trautmann2004a} as wells as a control
+chart for the desirability index with analysis of out-of
+control-signals are implemented \citep{trautmann2004}. The
+Desirability Pareto-Concept allows focussing on relevant parts of the
+Pareto-front by integrating a-priori-expert-knowledge in the
+multicriteria optimization process \citep{mehnen2007}.
+
+We will focus on the implementation of the Desirability Pareto-Concept
+in \textsf{R}. First we will give a short review of the traditional
+optimization strategy using desirability indices. Then, after
+showcasing NSGA-II \citep{dep2002}, we will briefly talk about how
+desirability functions can be integrated into optimization procedures
+that estimate the pareto front. Finally some of the problems faced
+during the development will be discussed. These include interfacing
+\textsf{R} and \textsf{C} code and using functions as first class objects.
+
+%% Kann man das so schreiben? Oder lieber ankündigen und ggf. unter
+%% den Tisch fallen lassen? Es ist ja jetzt doch wieder sehr viel für
+%% 15 min.
+In addition a short overview of the package will be given.
  
-One of the challenges faced when developing a new package is which implementation of
-object orientation to use. The S language and its implementation \textsf{R} provide two
-very different approaches. S3 methods are a straight forward implementation, where method
-dispatch done based on the class of one of the arguments. This is both fast and easy to
-implement. On the other hand, S4 provides a powerful method dispatch system and much nicer
-data encapsulation with pre- and postconditions as well as a certain degree of type
-safety. Some of the aspects we will cover in the talk include ease of implementation, the
-speed of method invocation and how these systems deal with functions as first class
-objects. Especially the last aspect will be dealt with, since function objects are a
-natural way to represent a desirability.
- 
-The last part of the talk will focus on ways to improve the speed of critical functions by
-rewriting them in \textsf{C(++)}.
- 
 \bibliography{useR_2008-abstract}{}
 \bibliographystyle{plainnat}
 \end{document}

