ExSTraCS – Extended Supervised Tracking and Classifying System

Ryan Urbanowicz is pleased to announce an advanced LCS for datamining:

 

This advanced machine learning algorithm is a Michigan-style learning classifier system (LCS) developed to specialize in classification, prediction, data mining, and knowledge discovery tasks. Michigan-style LCS algorithms constitute a unique class of algorithms that distribute learned patterns over a collaborative population of of individually interpretable IF:THEN rules, allowing them to flexibly and effectively describe complex and diverse problem spaces. ExSTraCS was primarily developed to address problems in epidemiological data mining to identify complex patterns relating predictive attributes in noisy datasets to disease phenotypes of interest. ExSTraCS combines a number of recent advancements into a single algorithmic platform. It can flexibly handle (1) discrete or continuous attributes, (2) missing data, (3) balanced or imbalanced datasets, and (4) binary or many classes. A complete users guide for ExSTraCS is included. Coded in Python 2.7.

 

[http://sourceforge.net/projects/exstracs/]

Posted in Uncategorized | Leave a comment

Educational LCS

Ryan Urbanowicz is pleased to announce the availability of an educational LCS.

The Educational Learning Classifier System (eLCS) is a set of code demos that are intended to serve as an educational resource to learn the basics of a Michigan-Style Learning Classifier System (modeled after the XCS and UCS algorithm architectures).  This resource includes 5 separate implementation demos that sequentially add major components of the algorithm in order to highlight how these components function, and how they are implemented into the algorithm code.

[http://sourceforge.net/projects/educationallcs/]

Posted in Software | Leave a comment

GAssist and BioHEL code resource

Jaume Bacardit has the source code for two excellent Pittsburgh style LCS for data mining tasks available at:

 

Interdisciplinary Computing and Complex BioSystems (ICOS) research group

 

[http://ico2s.org/]

Thank you to the project team for posting this code.

 

Posted in Software | Leave a comment

SIGEVOlution Volume 6, Issue 3-4, is now available

Looks like a great new issue, with:

  • Lunch Isn’t Free – But Cells Are by Moshe Sipper & Achiya Elyasaf
  • News from the GP Bibliography by William B. Langdon
  • Evostar Conference Report by Justyna Petke
  • Calls and Calendar


The newsletter is intended to be viewed electronically. Thanks to Pier Luca Lanzi, SIGEvolution Editor-in-Chief.

Comments Off

Call For Papers Special Session: Evolutionary Machine Learning for SEAL 2014

Call For Papers
Special Session: Evolutionary Machine Learning

The 10th International Conference on Simulated Evolution And Learning (SEAL 2014)
15-18 December 2014, Dunedin, New Zealand

http://seal2014.otago.ac.nz/

================

Machine learning and evolutionary computation are two major fields of computational intelligence. They share many fundamental similarities and are frequently explored together to tackle complex, large-scale, and dynamic learning problems under various sources of uncertainties.

This special session will cover a broad range of topics related to evolutionary machine learning, including novel learning algorithms and their innovative applications. We will focus on both theoretical and practical research in this field. The aim is to show how the global search performed by evolutionary methods can complement the local search of non-evolutionary methods and how the combination of the two can improve learning effectiveness and performance within a wide range of clustering, classification, regression, prediction, and control tasks.

Topics of interest include, but not limited to:

- Learning Classifier Systems

- Genetic Programming (GP) and its application to machine learning tasks

 - Evolutionary ensembles

 - Neuroevolution and its application to machine learning tasks

 - Genetic fuzzy systems

 - Hyper-parameter tuning with evolutionary methods

 - Theoretical analysis of evolutionary learning algorithms

 - Interesting practical applications

 - Advanced computing platforms for evolutionary machine learning

 - Other Genetics-Based Machine Learning: hybrid learning systems combining evolutionary techniques with machine learning methods

  ================
Important Dates:
28 July 2014, deadline for submission of full papers (<=12 pages)
29 August 2014, Notification of acceptance
16 September 2014, Deadline for camera-ready copies of accepted papers
15-18 December 2014, Conference sessions (including tutorials and workshops)
================
Paper Submission:
You should follow the SEAL 2014 Submission Web Site
(http://seal2014.otago.ac.nz/submissions.aspx). In the Main Research Topic, please choose
“Evolutionary Machine Learning ”
Special session papers are treated the same as regular conference papers. All papers will be fully refereed by a minimum of two specialized referees. Before final acceptance, all referees comments must be considered. All accepted papers that are presented at the conference will be included in the conference proceedings, to be published in Lecture Notes in Computer Science (LNCS) by Springer. Selected papers will be invited for further revision and extension for possible publication in a special issue of a SCI journal after further review Soft Computing (Springer, Impact Factor 1.124).
================
Special Session Organizers:
Dr Aaron Chen
School of Engineering and Computer Science, Victoria University of Wellington,
PO Box 600, Wellington, New Zealand.
Email: aaron.chen@ecs.vuw.ac.nz
Homepage: http://ecs.victoria.ac.nz/Main/AaronChen

Dr Will Browne
School of Engineering and Computer Science, Victoria University of Wellington,
PO Box 600, Wellington, New Zealand.
Email: will.browne@vuw.ac.nz
Homepage: http://ecs.victoria.ac.nz/Main/WillBrowne

Posted in announcements, Call for papers, Learning classifier systems | Leave a comment

Call For Papers Special Session: Evolutionary Feature Reduction The 10th International Conference on Simulated Evolution And Learning (SEAL 2014)

Call For Papers
Special Session: Evolutionary Feature Reduction
The 10th International Conference on Simulated Evolution And Learning (SEAL 2014)
15-18 December 2014, Dunedin, New Zealand

http://seal2014.otago.ac.nz/

================

Large numbers of features/attributes are often problematic in machine learning and data mining. They lead to conditions known as “the cures of dimensionality”. Feature reduction aims to solve this problem by selecting a small number of original features or constructing a smaller set of new features. Feature selection and construction are challenging tasks due to the large search space and feature interaction problems. Recently, there has been increasing interest in using evolutionary computation approaches to solve these problems.

The theme of this special session is the use of evolutionary computation for feature reduction, covering ALL different evolutionary computation paradigms including evolutionary algorithms, swarm intelligence, learning classifier systems, harmony search, artificial immune systems, and cross-fertilization of evolutionary computation and other techniques such as neural networks, and fuzzy and rough sets. This special session aims to investigate both the new theories and methods in different evolutionary computation paradigms to feature reduction, and the applications of evolutionary computation for feature reduction. Authors are invited to submit their original and unpublished work to this special session.
Topics of interest include but are not limited to:
• Feature ranking/weighting
• Feature subset selection
• Dimensionality reduction
• Feature construction
• Filter, wrapper, and embedded feature selection
• Hybrid feature selection
• Feature reduction for both supervised and unsupervised learning
• Multi-objective feature reduction
• Feature reduction with imbalanced data
• Analysis on evolutionary feature reduction methods
• Real-world applications of evolutionary feature reduction, e.g. gene analysis, bio-marker detection, et al.
================
Important Dates:
28 July 2014, deadline for submission of full papers (<=12 pages)
29 August 2014, Notification of acceptance
16 September 2014, Deadline for camera-ready copies of accepted papers
15-18 December 2014, Conference sessions (including tutorials and workshops)
================
Paper Submission:
You should follow the SEAL 2014 Submission Web Site
(http://seal2014.otago.ac.nz/submissions.aspx). In the Main Research Topic, please choose
“Evolutionary Feature Reduction”
Special session papers are treated the same as regular conference papers. All papers will be fully refereed by a minimum of two specialized referees. Before final acceptance, all referees comments must be considered. All accepted papers that are presented at the conference will be included in the conference proceedings, to be published in Lecture Notes in Computer Science (LNCS) by Springer. Selected papers will be invited for further revision and extension for possible publication in a special issue of a SCI journal after further review Soft Computing (Springer, Impact Factor 1.124).
================
Special Session Organizers:
Dr Bing Xue
School of Engineering and Computer Science, Victoria University of Wellington,
PO Box 600, Wellington, New Zealand.
Email: bing.xue@ecs.vuw.ac.nz
Homepage: http://ecs.victoria.ac.nz/Main/BingXue

Dr Kourosh Neshatian
Computer Science and Software Engineering, College of Engineering, University of Canterbury
Email: kourosh.neshatian@canterbury.ac.nz
Homepage: http://www.cosc.canterbury.ac.nz/kourosh.neshatian/
================
Note: Please reply to me if you want to be a Program Committee Member.

Best regards,
Bing

***************************************
Room CO 351, Cotton Building
Victoria University of Wellington
PO Box 600, Wellington 6140,
New Zealand
Mobile Phone; +64 220327481
Phone: +64-4-463 5233+ext 8874
Email: xuebingfifa@gmail.com
Bing.Xue@ecs.vuw.ac.nz
****************************************

Posted in Uncategorized | Leave a comment

Analyzing a Decade of Human-Competitive (“HUMIE”) Winners: What Can We Learn?

Many of us witnes the wonders of evolutionary computation (EC) on a daily basis as we put Darwin’s dangerous idea (a la Daniel Dennett) to good (safe) use. At the turn of the millennium John Koza noticed how EC had matured to the point of producing results that competed with humans. In 2004 John founded the “HUMIES” competition: Human-Competitive Results Produced by Genetic and Evolutionary Computation.

This annual competition has generated a lot of hubbub and — more importantly — tons of great results of both scientific and industrial value. Lee Spector, at Hampshire College, and I have been interested in the human-competitive angle of EC for several years. Between the two of us, we’re proud to be hugging eight HUMIE awards. Sadly, Lee has in the meantime been swayed by the Dark Side, joining the panel of judges for this illustrious award … May the force be with you, Lee :-)

Lee’s bright students, Karthik Kannappan, Tom Helmuth, Bill Lacava, Jake Wisdom, and Omri Bernstein joined the merry HUMIE bandwagon and we brainstormed on the merits of human-competitive machine evolution.

This past May we presented an analysis of a decade’s worth of HUMIE winners at the GPTP workshop in Ann Arbor, MI (thanks Rick Riolo, Bill Worzel, and Mark Kotanchek for organizing a wonderful event!).

An advanced draft of our paper is now available for download: Analyzing a Decade of Human-Competitive (“HUMIE”) Winners: What Can We Learn?

Enjoy!

moshe sipper, writer & professor

Comments Off

On the Landscape of Combinatorial Optimization Problems

This paper carries out a comparison of the fitness landscape for four classic optimization problems: Max-Sat, graph-coloring, traveling salesman, and quadratic assignment. We have focused on two types of properties, local average properties of the landscape, and properties of the local optima. For the local optima we give a fairly comprehensive description of the properties, including the expected time to reach a local optimum, the number of local optima at different cost levels, the distance between optima, and the expected probability of reaching the optima. Principle component analysis is used to understand the correlations between the local optima. Most of the properties that we examine have not been studied previously, particularly those concerned with properties of the local optima. We compare and contrast the behavior of the four different problems. Although the problems are very different at the low level, many of the long-range properties exhibit a remarkable degree of similarity.

Comments Off

IEEE Transactions on Evolutionary Computation information for authors

Provides instructions and guidelines to prospective authors who wish to submit manuscripts.

Comments Off

Optimal Experiment Design for Coevolutionary Active Learning

This paper presents a policy for selecting the most informative individuals in a teacher-learner type coevolution. We propose the use of the surprisal of the mean, based on Shannon information theory, which best disambiguates a collection of arbitrary and competing models based solely on their predictions. This policy is demonstrated within an iterative coevolutionary framework consisting of symbolic regression for model inference and a genetic algorithm for optimal experiment design. Complex symbolic expressions are reliably inferred using fewer than 32 observations. The policy requires 21% fewer experiments for model inference compared to the baselines and is particularly effective in the presence of noise corruption, local information content as well as high dimensional systems. Furthermore, the policy was applied in a real-world setting to model concrete compression strength, where it was able to achieve 96.1% of the passive machine learning baseline performance with only 16.6% of the data.

Comments Off