IlliGAL Blogging

In yesterday's main Gamasutra technical feature, Rick Strom explored the possibility of improving in-game AI by evolving pathfinding algorithms using genetic programming. More information can be found here.

Web wire reports that MWH Soft released "first-of-its-kind InfoWater VCM, The Ultimate Solution for Infrastructure Criticality and Vulnerability Assessment". The software contains a genetic algorithm as one of its components:

Built atop ArcGIS (ESRI, Redlands, CA) and drawing on the most advanced numerical computation and genetic algorithms optimization technologies, Infowater effortlessly reads GIS datasets; corrects network topology problems and data flaws; extracts pertinent modeling information; and automatically constructs, skeletonizes, loads, calibrates and generates optimized design, rehabilitation and pump scheduling solutions with astounding speed.

A related post can be found here.

The Missouri Estimation of Distribution Algorithms Laboratory (MEDAL) is pleased to announce publication of the following MEDAL technical reports:

MEDAL Report No. 2006007
Substructural Neighborhoods for Local Search in the Bayesian Optimization Algorithm
Claudio F. Lima, Martin Pelikan, Kumara Sastry, Martin Butz, David E. Goldberg, and Fernando G. Lobo (2006)
Also IlliGAL Report No. 2006021.

MEDAL Report No. 2006006
Population Sizing for Entropy-based Model Building in Genetic Algorithms
Tian-Li Yu, Kumara Sastry, David E. Goldberg, and Martin Pelikan (2006)
Also IlliGAL Report No. 2006020.

MEDAL Report No. 2006005
Order or Not: Does Parallelization of Model Building in hBOA Affect Its Scalability?
Martin Pelikan and James D. Laury, Jr. (2006)

The above and other MEDAL technical reports can be found here.

The workshop Optimization by Building and Using Probabilistic Models (OBUPM-2006) will take place at the Genetic and Evolutionary Computation Conference (GECCO-2006) in Seattle, WA. OBUPM-2006 has been scheduled for the afternoon session of GECCO-2006 on Sunday, July 9, 2006. The workshop is organized by Peter A.N. Bosman, Joern Grahl, Kumara Sastry and Martin Pelikan.

The list of OBUPM-2006 presentations follows:

• Peter A.N. Bosman
  Matching Inductive Search Bias and Problem Structure in Continuous Estimation of Distribution Algorithms
  
  
• Joern Grahl
  Additively Decomposable Real-World Problems: General Findings and Examples from Logistics
  
  
• Tian-Li Yu
  Population Sizing for Entropy-based Model Building in Genetic Algorithm
  
  
• Claudio Lima
  Substructural Neighborhoods for Local Search in the Bayesian Optimization Algorithm
  
  
• Xavier Llora
  Chi-ary Extended Compact Classifier Systems: Linkage Learning in Pittsburgh LCS
  
  
• Moshe Looks, Ben Goertzel
  Probabilistic Model-Building for Program Learning: The Challenge and Opportunity of a Complex Representation
  

OBUPM-2006 is open to all GECCO-2006 attendees.

The preliminary schedule for the GECCO conference (Genetic and Evolutionary Computation Conference) is posted here and it looks like another winner. Even years are typically slow ones, but conference registration is ahead of plans and it looks as though attendance will keep pace with recent conferences. More information about registration can be found here.

The special issue of the International Journal of Computational Intelligence Research (IJCIR) with the focus on evolutionary multiobjective optimization is now available online here. The special issue has been edited by Marco Laumanns, Sanaz Mostaghim, Günter Rudolph, and Jürgen Teich.

at TEE here.

Bruker Daltonics ClinProTools 2.1 uses genetic algorithms and other techniques for biomarker panel analysis, biofluid profiling, multivariate data analysis of large sample cohorts, classical statistics and sample classification (see here):

ClinProTools 2.1 now includes many additional features for data analysis and visualization: the new proprietary Supervised Neural Network(TM) algorithm allows a third independent multiclass, multivariate analysis approach, complementing the company's existing Support Vector Machine and Modified Genetic Algorithm. Furthermore, it supports univariate peak statistics and a proprietary QuickClassifier (TM) algorithm. In addition to these supervised approaches, ClinProTools 2.1 also includes Principle Component Analysis (PCA) for unsupervised data analysis.

More information is available at the company's web site (here).

For a paper on using genetic algorithms to simulate rodent populations, see here.