1/4/2012 - By creating a model in Testing Mode, you have more control over the quality of the model at making predictions. Now in GeneXproTools 4.3, by using the Import Models feature, it’s possible to create a model while in Testing Mode and then import it from a different run primed for prediction. This video shows how to do this in GeneXproTools using the monthly closings of the Dow-Jones industrial index. It also shows the difference between Testing Mode and Prediction Mode in GeneXproTools using a very simple time series example.

To watch the video, please go to:

http://www.gepsoft.com/videos/TimeSeriesPredictionInTestingMode.htm

To download the Demo of GeneXproTools, please go to:

http://www.gepsoft.com/Downloads.htm

The demo allows you to use your own time series and all the features are operational, except for making predictions and code visualization. However, by evaluating the software in Testing Mode, you have access to all the model statistics to give you an idea of the quality of the models being created.

12/1/2011 - A new video showing how to do Logistic Regression and Classification in GeneXproTools, focusing on the new features introduced in version 4.3.

The highlights are:

o How to Create a New Modeling Session
o How to Create Independent and Unattended Runs
o How to Import Models from Different Files
o How to Convert Logistic Regression Models to Classification and Vice Versa
o New Logistic Regression Charts
o New Binomial Fit by Model Chart
o New Binomial Fit by Target Chart
o New Classification Tapestry
o New Model Management Tools
o How to Create Ensemble Models and Deploy them to Excel
o How to Deploy Models to Excel

To watch the video, please go to:

http://www.gepsoft.com/videos/WhatsNewGeneXproTools43LogisticRegressionClassification.htm

To download the Demo of GeneXproTools, please go to:

http://www.gepsoft.com/Downloads.htm

The demo allows you to use your own datasets and all the features are operational, except for scoring/prediction and code visualization. However, for the sample runs that ship with GeneXproTools, all the features are totally functional. The sample runs cover a wide spectrum of real-world modeling problems, from credit approval to diabetes and cancer diagnosis.

For a more detailed list of the new features introduced in all modeling platforms of GeneXproTools, please go to:

http://www.gepsoft.com/WhatsNew.htm

11/30/2011 - We are proud to announce that Gepsoft released today GeneXproTools 4.3. With this release we are introducing new visualization tools, including a Classification Tapestry for model visualization during the design process and Binomial Fit Charts for visualization of Logistic Regression Models.

We are also adding new possibilities with the deployment of models to Excel that will save you the time and resources of dealing with the model source code. In addition to Model Deployment to Excel, we are also introducing the deployment of ensemble models to Excel, with the automatic generation of the Majority Vote Model for Classification and Logistic Regression, and the Average and Median Models for Function Finding and Time Series Prediction.

Another important aspect of this release is the introduction of new Fitness Functions for exploring the solution space more efficiently, creating better classifiers with wider margins and better generalization.

Also worth pointing out are the new capabilities for multiple unattended runs and the extensive Model Management Tools for selecting and keeping only the models of interest.

Below is a more detailed list of the new features:

o New Visualization Tools
o Model Deployment to Excel
o Ensemble Deployment to Excel
o Extensive Model Management
o New Fitness Functions
o Speed-ups
o Decentralized Model Navigation
o Improved Confusion Matrix
o New Defaults for the Genetic Operators
o Real-time Evaluation of the Correlation Coefficient
o Change Rate Indicator
o Improved Heatmap
o Improved Evolutionary Dynamics Chart
o Improved Addition of Neutral Genes
o New Selection Tools for the Function Set
o Better Handling of Big Function Sets
o New Sorting Tools in the Results Panel
o New Copy Chart Data
o New Copy Modes for the Results Table
o New Custom Fitness Functions Examples
o New UDFs Examples
o New Sample Runs
o Improved Defaults
o Improved History Panel
o Copy of the Confusion Matrix
o Scoring Models in Excel
o Support for Excel 2007 and 2010
o Easier Data Loading in Excel
o Drag and Drop of GEP Files
o More Compact GEP Files
o And other little improvements that will make modeling more enjoyable.

You can see a more detailed list of the new features here:

http://www.gepsoft.com/WhatsNew.htm

GeneXproTools ships with 20 different sample runs that cover a large array of modeling problems. The sample runs are unlocked in demo mode so you can create new models, generate code and generally evaluate the complete workflow.

The demo allows you to use your own datasets and all the features are operational except scoring/prediction and code visualization. However, for the included sample runs all the features are operational in the demo. GeneXproTools can be installed on Windows XP, Windows 7, and Windows Server 2003 and 2008. To download GeneXproTools, please go to:

http://www.gepsoft.com/Downloads.htm

11/7/2010 - Launched September 2010 by Jason Thomas, the GEP4J project is an open-source implementation of Gene Expression Programming in Java. From the project summary: “This project is in the early phases, but you can already do useful things such as evolving decision trees (nominal, numeric, or mixed attributes) with ADF's (automatically defined functions), and evolve functions.” GEP4J is available from Google Project Hosting:

http://code.google.com/p/gep4j/

5/15/2008 - The GEP Online Bibliography has a new host (Ryerson University, Canada) and a new system. The system, created by Marcus Vinicius dos Santos, allows the authors themselves to add new entries to the Biblio. No usernames or passwords are necessary and the new entries/edits become immediately available, which makes everything much more enjoyable.

The GEP Online Bibliography is a great place for authors to make their work visible and accessible to others. So if you have any GEP-related work or know of any other publications by other researchers that are not yet on the GEP Biblio, please come join us and add them to the list:

http://pliant.scs.ryerson.ca/gepbiblio/

Thank you for your contribution!

3/25/2008 - We are proud to announce the addition of the Logistic Regression Analytics Platform to GeneXproTools 4.0. This new and exciting technology merges the modeling power of GeneXproTools with a set of standard statistical analyses to help you create ranking systems and calculate probabilities for your model scores. The Logistic Regression Analytics Platform includes Quantile Analysis and Regression, ROC Curves, Cutoff Points, Gains and Lift Charts, Logistic Regression and Logistic Fit, and Logistic and ROC Confusion Matrixes.

This is a free upgrade for licensed users of GeneXproTools 4.0 and the only action you need to take is to download the new version, uninstall your current copy and run the new setup file. The new installation file can be downloaded from the downloads page.

There is also a new release of GeneXproServer and licensed users will receive shortly a new link for downloading the new installation file.

11/27/2007 - This second maintenance release of GeneXproTools 4.0 addresses minor issues in the calculation of the fitness functions improving their robustness against overflows. It also includes a few minor bug fixes and improvements. We recommend that you download the setup file, uninstall GeneXproTools and install it again using this newer setup file. You can download the new file from the Gepsoft website:

http://www.gepsoft.com/downloads.htm

We are also releasing today the first maintenance release for GeneXproServer 1.0. This new release includes all the optimizations implemented in the two previous updates of GeneXproTools 4.0, including the important modifications implemented to reduce substantially the memory footprint of the software, thus allowing the use of considerably larger datasets. All GeneXproServer users will be contacted directly by email with their download information, but if you missed it don’t hesitate to request it from us.

8/21/2007 - The invited book chapter on Automatically Defined Functions in GEP is now available online both in pdf format and html at:

http://www.gene-expression-programming.com/webpapers/abstracts.asp#13

Ferreira, C., Automatically Defined Functions in Gene Expression Programming. In N. Nedjah, L. de M. Mourelle, A. Abraham, eds., Genetic Systems Programming: Theory and Experiences, Studies in Computational Intelligence, Vol. 13, pp. 21-56, Springer-Verlag, 2006.

ABSTRACT: In this chapter it is shown how Automatically Defined Functions are encoded in the genotype/phenotype system of Gene Expression Programming. As an introduction, the fundamental differences between Gene Expression Programming and its predecessors, Genetic Algorithms and Genetic Programming, are briefly summarized so that the evolutionary advantages of Gene Expression Programming are better understood. The introduction proceeds with a detailed description of the architecture of the main players of Gene Expression Programming (chromosomes and expression trees), focusing mainly on the interactions between them and how the simple, yet revolutionary, structure of the chromosomes allows the efficient, unconstrained exploration of the search space. The work proceeds with an introduction to Automatically Defined Functions and how they are implemented in Gene Expression Programming. Furthermore, the importance of Automatically Defined Functions in Evolutionary Computation is thoroughly analyzed by comparing the performance of sophisticated learning systems with Automatically Defined Functions with much simpler ones on the sextic polynomial problem.

8/14/2007 - The paper on complete neural network induction using Darwinian evolution is now available online both in pdf format and html at:

http://www.gene-expression-programming.com/webpapers/abstracts.asp#14

Ferreira, C., Designing Neural Networks Using Gene Expression Programming. In A. Abraham, B. de Baets, M. Köppen, and B. Nickolay, eds., Applied Soft Computing Technologies: The Challenge of Complexity, pages 517-536, Springer-Verlag, 2006.

ABSTRACT: An artificial neural network with all its elements is a rather complex structure, not easily constructed and/or trained to perform a particular task. Consequently, several researchers used genetic algorithms to evolve partial aspects of neural networks, such as the weights, the thresholds, and the network architecture. Indeed, over the last decade many systems have been developed that perform total network induction. In this work it is shown how the chromosomes of Gene Expression Programming can be modified so that a complete neural network, including the architecture, the weights and thresholds, could be totally encoded in a linear chromosome. It is also shown how this chromosomal organization allows the training/adaptation of the network using the evolutionary mechanisms of selection and modification, thus providing an approach to the automatic design of neural networks. The workings and performance of this new algorithm are tested on the 6-multiplexer and on the classical exclusive-or problems.

This paper requires a certain familiarity with the basics of GEP, especially the head/tail organization, the expression of genes with random constants, and the type and mechanisms of the genetic operators. For a quick introduction see my Complex Systems paper.

For the sample problems of this paper I chose well-known logical functions, but the beauty of GEP-nets is that they can be used on a multitude of modeling problems, from nonlinear regression to classification and they are as good as any GEP system. I guess I’ll have to write a paper on this since I haven’t seen anyone taking up on this task since I first described this algorithm in my 2002 book.

8/7/2007 - The new videos are quick tours to the four modeling categories of GeneXproTools and are accessible from here either to download or watch online:

http://www.gepsoft.com/videos.htm

An email interview conducted by Kim Patch for a story on GEP for Technology Research News in 2001 is now available here:

http://www.gene-expression-programming.com/Interview2001.asp
 

7/31/2007 - This new release reduces the memory footprint of GeneXproTools and therefore allows the use of considerably larger datasets. It also adds support for Unix-style text data files and includes a few minor bug fixes and improvements.

To download this new release of GeneXproTools 4.0, please go to:

http://www.gepsoft.com/downloads.htm.

You will need to uninstall GeneXproTools 4.0 and reinstall this new version.

7/19/2007 - The tutorial External Custom Fitness has been updated with new projects in VB.NET and C# and a new free library that provides direct access to the datasets inside the GeneXproTools run.

The tutorial files now include a complete VB6 project, a solution with two projects (one in C# and another in VB.NET) and the GXPT4CFHelper library.

The Custom Fitness Function is one of the most popular features in GeneXproTools and has been explored by a significant number of users. And again and again we were asked to allow total access to the datasets. This is now supported using this new library. Enjoy!

The full tutorial and all the required files can be found at the Gepsoft website:

http://www.gepsoft.com/tutorial004.htm

4/25/2007 - Ryan O'Neil, a graduate student from George Mason University, released into the open source community his GEP library for Python. In his words, "PyGEP is a simple library suitable for academic study of Gene Expression Programming in Python 2.5, aiming for ease of use and rapid implementation. It provides standard multigenic chromosomes; a population class using elitism and fitness scaling for selection; mutation, crossover and transposition operators; and some standard GEP functions and linkers." PyGEP is hosted at http://code.google.com/p/pygep/.

4/18/2007 - Gepsoft launched today its Associates Program with the aim of creating a wider web of GeneXproTools users. Joining the program is easy and very rewarding as associates earn 10 percent in referral fees. As an associate, all you need is to link to Gepsoft.com from your website or blog through specially formatted links. These links as well as animated gifs and banners are accessible to associates through Gepsoft’s exclusive extranet. With them you can create appealing webpages to drive traffic to Gepsoft’s website.

For more information click here.

3/12/2007 - I compiled some of the questions I answered over the years about GEP through email. Answering all of them took time and effort and I thought it would be great if others could learn from them. They span quite a few years and cover a lot of ground, but each one of them is self-contained and hopefully you’ll find them also of interest to you. For privacy sake, I removed any references that could identify the people involved. They can be found at:

http://www.gene-expression-programming.com/Q&A.asp

-- Candida Ferreira

1/30/2007 - The new GeneXproTools FAQ was compiled from support requests received over the years, and covers Installation Issues and Demo Functionality, Classification and General Issues, plus Algorithms and Evolution Issues. Thanks to everyone who contributed!

Three new tutorials:

1. DNA Microarrays: A Case Study
2. What is GEP?
3. Karva Notation: The Native Language of GeneXproTools

These new tutorials are the first three of what I hope will grow into an interesting forum not only to discuss practical and theoretical matters but also for talking about different uses of GeneXproTools. There are people around the world using GeneXproTools to study amazing things (from top quark and Higgs analysis to new drugs and analog circuits) and it would be great to have them using this forum to talk about their work and what they are achieving with GeneXproTools (I encourage and invite you to submit your contributions to me personally).

-- Candida Ferreira

9/27/2006 - Gepsoft is releasing today the software GeneXproServer 1.0, an add-on to GeneXproTools 4.0 for integrated modeling.

GeneXproServer is a batch processor of gep runs that can be easily integrated with any infrastructure and process to enable the discovery and management of hundreds of models over different datasets and generated under different settings.

GeneXproServer is shipped in both Windows and console applications. The former is appropriate for desktop computers and for short jobs, whereas the latter can be easily integrated with other systems and workflows. Either mode provides facilities to trigger external applications at key points during the job.

To know more visit the GeneXproServer page at:

http://www.gepsoft.com/gxps.htm

9/23/2006 - An Interview with Cândida Ferreira on Gene Expression Programming (in Portuguese). By Ricardo Carvalho for ALAGAMARES (23.09.2006).

7/17/2006 - Gepsoft is releasing today GeneXproTools 4.0, the new name and version of Gepsoft APS.

GeneXproTools is a data analysis tool that creates predictive models from raw numerical data. The main advantages of GeneXproTools are its user friendliness and its powerful algorithms. The main focus of GeneXproTools is a streamlined process that starts with the loading of the data, progresses to the automatic creation of the model and concludes either with the conversion of the model into any of the 16 supported languages or with the scoring of external data.

GeneXproTools 4.0 is packed with several new features and improvements and a redesign of the user interface. The main new features are:

o Unlimited data samples for all editions
o A new edition with unlimited variables (Enterprise Edition)
o A new Academic edition also with unlimited variables (Academic Enterprise)
o Much faster data loading and smaller run files
o Model simplification algorithms
o A new algorithm for Logic Synthesis
o Several new fitness functions
o A very fast Custom Fitness implementation
o Support for many more programming languages (16 in total)
o More than 200 new functions
o New Run Panel with six new charts including program size and real-time curve fitting

For a more detailed list of the new features please go to:

http://www.gepsoft.com/WhatsNew.htm


The new Enterprise Edition with unlimited variables opens new and exciting possibilities in fields like DNA microarrays. For a simple tutorial on DNA microarrays using the well-known ALL-AML leukemia data with 7,129 variables please go to:

http://www.gepsoft.com/Article001.htm


GeneXproTools ships with 16 different sample runs that cover a large array of modeling problems, including the DNA microarray problem with 7,129 variables. The sample runs are unlocked in demo mode so you can create new models, generate code and generally evaluate the complete workflow.

The demo allows you to use your own datasets and all the features are operational except scoring/prediction and visualization of the models. However, for the included sample runs all the features are operational in the demo. GeneXproTools can be installed on Windows 2000, XP or Server 2003. To download GeneXproTools, please go to:

http://www.gepsoft.com/Downloads.htm

5/29/2006 - The second, substantially extended and revised edition of the GEP book is now available!

Gene Expression Programming: Mathematical Modeling by an Artificial Intelligence by Candida Ferreira
2nd edition 2006,
478 p. 159 illus., Hardcover
Springer-Verlag, Germany
ISBN: 3-540-32796-7
More...

About the book:

Candida Ferreira thoroughly describes the basic ideas of gene expression programming (GEP) and numerous modifications to this powerful new algorithm. This monograph provides all the implementation details of GEP so that anyone with elementary programming skills will be able to implement it themselves. The book also includes a self-contained introduction to this new exciting field of computational intelligence, including several new algorithms for decision tree induction, data mining, classifier systems, function finding, polynomial induction, times series prediction, evolution of linking functions, automatically defined functions, parameter optimization, logic synthesis, combinatorial optimization, and complete neural network induction. The book also discusses some important and controversial evolutionary topics that might be refreshing to both evolutionary computer scientists and biologists.

This second edition has been substantially revised and extended with five new chapters, including a new chapter describing two new algorithms for inducing decision trees with nominal and numeric/mixed attributes.

Book Chapters:

1 Introduction: The Biological Perspective
2 The Entities of Gene Expression Programming
3 The Basic Gene Expression Algorithm
4 The Basic GEA in Problem Solving
5 Numerical Constants and the GEP-RNC Algorithm
6 Automatically Defined Functions in Problem Solving
7 Polynomial Induction and Time Series Prediction
8 Parameter Optimization
9 Decision Tree Induction
10 Design of Neural Networks
11 Combinatorial Optimization
12 Evolutionary Studies

1/30/2006 - The invited book chapter published in Studies in Computational Intelligence is now available online:

http://www.gene-expression-programming.com/webpapers/abstracts.asp#13

Ferreira, C., Automatically Defined Functions in Gene Expression Programming. In N. Nedjah, L. de M. Mourelle, A. Abraham, eds., Genetic Systems Programming: Theory and Experiences, Studies in Computational Intelligence, Vol. 13, pp. 21-56, Springer-Verlag, 2006.

ABSTRACT:
In this chapter it is shown how Automatically Defined Functions are encoded in the genotype/phenotype system of Gene Expression Programming. As an introduction, the fundamental differences between Gene Expression Programming and its predecessors, Genetic Algorithms and Genetic Programming, are briefly summarized so that the evolutionary advantages of Gene Expression Programming are better understood. The introduction proceeds with a detailed description of the architecture of the main players of Gene Expression Programming (chromosomes and expression trees), focusing mainly on the interactions between them and how the simple, yet revolutionary, structure of the chromosomes allows the efficient, unconstrained exploration of the search space. The work proceeds with an introduction to Automatically Defined Functions and how they are implemented in Gene Expression Programming. Furthermore, the importance of Automatically Defined Functions in Evolutionary Computation is thoroughly analyzed by comparing the performance of sophisticated learning systems with Automatically Defined Functions with much simpler ones on the sextic polynomial problem.

8/21/2005 - The invited book chapter published by Idea Group Publishing is now available online both in pdf format and html at:

http://www.gene-expression-programming.com/webpapers/abstracts.asp#11

Ferreira, C., Gene Expression Programming and the Evolution of Computer Programs. In Leandro N. de Castro and Fernando J. Von Zuben, eds., Recent Developments in Biologically Inspired Computing, pages 82-103, Idea Group Publishing, 2004.

ABSTRACT: In this chapter an artificial problem solver inspired in natural genotype/phenotype systems – gene expression programming – is presented. As an introduction, the fundamental differences between gene expression programming and its predecessors, genetic algorithms and genetic programming, are briefly summarized so that the evolutionary advantages of gene expression programming are better understood. The work proceeds with a detailed description of the architecture of the main players of this new algorithm (chromosomes and expression trees), focusing mainly on the interactions between them and how the simple yet revolutionary structure of the chromosomes allows the efficient, unconstrained exploration of the search space. And finally, the chapter closes with an advanced application in which gene expression programming is used to evolve computer programs for diagnosing breast cancer.

4/28/2005 - The seventh and last chapter of the book GENE EXPRESSION PROGRAMMING: MATHEMATICAL MODELING BY AN ARTIFICIAL INTELLIGENCE is now available to browse online at:

http://www.gene-expression-programming.com/Books/

All the seven chapters and are now available online (1 Introduction; 2 The Entities of Gene Expression Programming; 3 The Basic Gene Expression Algorithm; and 4 The Basic GEA in Problem Solving; 5 Design of Neural Networks; 6 Combinatorial Optimization; and 7 Evolutionary Studies).

Entries for Chapter 7:

Chapter 7: EVOLUTIONARY STUDIES

1. Genetic operators and their power
1.1. Comparing the performance of mutation, transposition, and recombination
1.2. Evolutionary dynamics of different types of GEP populations
1.2.1. Mutation
1.2.2. Transposition
1.2.3. Recombination
2. The founder effect
2.1. Choosing non-homogenizing and homogenizing populations to study the founder effect
2.2. Analyzing the founder effect in simulated evolutionary processes
3. Testing the building block hypothesis
4. The role of neutrality in evolution
4.1. Genetic neutrality in unigenic systems
4.2. Genetic neutrality in multigenic systems
5. The higher hierarchical organization of multigenic systems
6. The open-ended evolution of GEP populations
7. Analysis of different selection schemes

2/19/2005 - The sixth chapter of the book GENE EXPRESSION PROGRAMMING: MATHEMATICAL MODELING BY AN ARTIFICIAL INTELLIGENCE is now available to browse online at:

http://www.gene-expression-programming.com/Books/

The book has 7 chapters and all but the last are now available (1 Introduction; 2 The Entities of Gene Expression Programming; 3 The Basic Gene Expression Algorithm; and 4 The Basic GEA in Problem Solving; 5 Design of Neural Networks; and 6 Combinatorial Optimization). The last chapter (7 Evolutionary Studies) will follow shortly. Entries for Chapter 6:

Chapter 6: COMBINATORIAL OPTIMIZATION

1. Multigene families and scheduling problems
2. Combinatorial-specific operators: Performance and mechanisms
2.1. Inversion
2.2. Gene deletion/insertion
2.3. Restricted permutation
2.4. Other search operators
2.4.1. Sequence deletion/insertion
2.4.2. Generalized permutation
3. Two scheduling problems
3.1. The traveling salesperson problem
3.2. The task assignment problem
4. Evolutionary dynamics of simple GEP systems

12/18/2004 - The fifth chapter of the book GENE EXPRESSION PROGRAMMING: MATHEMATICAL MODELING BY AN ARTIFICIAL INTELLIGENCE is now available to browse online at:

http://www.gene-expression-programming.com/Books/index.asp

The book has 7 chapters and the first five are now available (1 Introduction; 2 The Entities of Gene Expression Programming; 3 The Basic Gene Expression Algorithm; and 4 The Basic GEA in Problem Solving; 5 Design of Neural Networks). The remaining two chapters (6 Combinatorial Optimization; and 7 Evolutionary Studies) will follow shortly.

Entries for Chapter 5:

Chapter 5: DESIGN OF NEURAL NETWORKS
1. Genes with multiple domains for neural network simulation
2. Special search operators
2.1. Domain-specific transposition
2.2. Intragenic two-point recombination
2.3. Direct mutation of weights and thresholds
3. Solving problems with GEP neural networks
3.1. Neural network for the exclusive-or problem
3.2. Neural network for the 6-multiplexer
4. Evolutionary dynamics of GEP-nets

10/23/2004 - The fourth chapter of the book GENE EXPRESSION PROGRAMMING: MATHEMATICAL MODELING BY AN ARTIFICIAL INTELLIGENCE is now available to browse online at:

http://www.gene-expression-programming.com/Books/index.asp

The book has 7 chapters and the first four are now available (1 Introduction; 2 The Entities of Gene Expression Programming; 3 The Basic Gene Expression Algorithm; and 4 The Basic GEA in Problem Solving). The remaining chapters (5 Design of Neural Networks; 6 Combinatorial Optimization; and 7 Evolutionary Studies) will follow shortly.

Entries for Chapter 4:

Chapter 4: THE BASIC GEA IN PROBLEM SOLVING
1. Symbolic regression
1.1. Function finding on a one-dimensional parameter space
1.2. Function finding on a five-dimensional parameter space
1.3. Mining meaningful information from noisy data
2. Symbolic regression and the creation of numerical constants
2.1. Manipulation of numerical constants in GEP
2.2. Two approaches to the problem of constant creation
2.2.1. Direct manipulation of numerical constants
2.2.2. Creation of numerical constants from scratch
3. Parameter optimization
3.1. Multigenic chromosomes and multidimensional parameter optimization
3.2. Maximum seeking with GEP
4. Time series prediction
4.1. Evolution of Kolmogorov-Gabor polynomials
4.2. Simulating STROGANOFF and enhanced STROGANOFF with GEP
4.3. Predicting sunspots with GEP
5. Classification problems
5.1. Diagnosis of breast cancer
5.2. Credit screening
5.3. Fisher’s irises
6. Logic synthesis
6.1. Finding solutions to odd-parity functions with the basic gene expression algorithm
6.2. Finding solutions to odd-parity functions with UDFs
6.3. Finding solutions to odd-parity functions with ADFs
7. Evolving cellular automata rules for the density-classification problem
7.1. The density-classification task
7.2. Two new rules discovered by GEP

10/11/2004 - The paper presented at the 9th Online World Conference on Soft Computing in Industrial Applications is now available online at:

http://www.gene-expression-programming.com/webpapers/abstracts.asp#12

Ferreira, C., Designing Neural Networks Using Gene Expression Programming. 9th Online World Conference on Soft Computing in Industrial Applications, September 20 - October 8, 2004.

ABSTRACT: An artificial neural network with all its elements is a rather complex structure, not easily constructed and/or trained to perform a particular task. Consequently, several researchers used Genetic Algorithms to evolve partial aspects of neural networks, such as the weights, the thresholds, and the network architecture. Indeed, over the last decade many systems have been developed that perform total network induction. In this work it is shown how the chromosomes of Gene Expression Programming can be modified so that a complete neural network, including the architecture, the weights and thresholds, could be totally encoded in a linear chromosome. It is also shown how this chromosomal organization allows the training/adaptation of the network using the evolutionary mechanisms of selection and modification, thus providing an approach to the automatic design of neural networks. The workings and performance of this new algorithm are tested on the 6-multiplexer and on the classical exclusive-or problems.

8/25/2004 - The third chapter of the book GENE EXPRESSION PROGRAMMING: MATHEMATICAL MODELING BY AN ARTIFICIAL INTELLIGENCE is now available to browse online at:

http://www.gene-expression-programming.com/Books/index.asp

The book has 7 chapters and the first three are now available (1 Introduction; 2 The Entities of Gene Expression Programming; and 3 The Basic Gene Expression Algorithm). The remaining chapters (4 The Basic GEA in Problem Solving; 5 Design of Neural Networks; 6 Combinatorial Optimization; and 7 Evolutionary Studies) will follow shortly.

Entries for Chapter 3:

Chapter 3: THE BASIC GENE EXPRESSION ALGORITHM
1. Populations of individuals
1.1. Creation of the initial population
1.2. Subsequent generations and elitism
2. Fitness functions and selection
2.1. Fitness functions and the selection environment
2.2. Selection
3. Reproduction with modification
3.1. Replication and selection
3.2. Mutation
3.3. Transposition and insertion sequence elements
3.3.1. Transposition of insertion sequence elements
3.3.2. Root transposition
3.3.3. Gene transposition
3.4. Recombination
3.4.1. One-point recombination
3.4.2. Two-point recombination
3.4.3. Gene recombination
4. Solving a simple problem with GEP

8/16/2004 - The paper published by Idea Group Publishing is now available to order online. For more information, please go to:

http://www.gene-expression-programming.com/author.asp

Ferreira, C., Gene Expression Programming and the Evolution of Computer Programs. In Leandro N. de Castro and Fernando J. Von Zuben, eds., Recent Developments in Biologically Inspired Computing, pages 82-103, Idea Group Publishing, 2004.

ABSTRACT:
In this chapter an artificial problem solver inspired in natural genotype/phenotype systems – gene expression programming – is presented. As an introduction, the fundamental differences between gene expression programming and its predecessors, genetic algorithms and genetic programming, are briefly summarized so that the evolutionary advantages of gene expression programming are better understood. The work proceeds with a detailed description of the architecture of the main players of this new algorithm (chromosomes and expression trees), focusing mainly on the interactions between them and how the simple yet revolutionary structure of the chromosomes allows the efficient, unconstrained exploration of the search space. And finally, the chapter closes with an advanced application in which gene expression programming is used to evolve computer programs for diagnosing breast cancer.

6/19/2004 - The second chapter of the book GENE EXPRESSION PROGRAMMING: MATHEMATICAL MODELING BY AN ARTIFICIAL INTELLIGENCE is now available to browse online at:

http://www.gene-expression-programming.com/Books/index.asp

This chapter describes the main players of GEP (chromosomes and expression trees) and introduces a varied set of chromosomal architectures including the encoding of ADFs and neural networks.

Entries for Chapter 2:

Chapter 2: THE ENTITIES OF GENE EXPRESSION PROGRAMMING
1. The genome
1.1. Open reading frames and genes
1.2. Structural and functional organization of genes
1.3. Multigenic chromosomes
1.4. Structural and functional diversity of chromosomes
2. Expression trees and the phenotype
2.1. Information decoding: Translation
2.2. Posttranslational interactions and linking functions
2.3. Cells and the evolution of linking functions
2.4. Other levels of complexity
2.5. Karva language: The language of GEP

6/6/2004 - The first chapter of the book GENE EXPRESSION PROGRAMMING: MATHEMATICAL MODELING BY AN ARTIFICIAL INTELLIGENCE is now available to browse online at:

http://www.gene-expression-programming.com/Books/index.asp

The book has 7 chapters (1 Introduction; 2 The Entities of Gene Expression Programming; 3 The Basic Gene Expression Algorithm; 4 The Basic GEA in Problem Solving; 5 Design of Neural Networks; 6 Combinatorial Optimization; and 7 Evolutionary Studies) and hopefully the remaining chapters will be posted over the next few months.

Entries for Chapter 1:

Chapter 1: INTRODUCTION
1. The entities of biological gene expression
1.1. DNA
1.2. RNA
1.3. Proteins
2. Biological gene expression
2.1. Genome replication
2.2. Genome restructuring: Mutation, recombination, transposition, and gene duplication 
2.2.1. Mutation
2.2.2. Recombination
2.2.3. Transposition
2.2.4. Gene duplication
2.3. Transcription
2.4. Translation and posttranslational modifications
2.4.1. Translation 
2.4.2. Posttranslational modifications
3. Adaptation and evolution
4. Genetic algorithms
5. Genetic programming
6. Gene expression programming

4/11/2004 - With this new service from Gepsoft you can create your own models using the Demo Version of APS 3.0 and then extract the best model to a special file format using the software APS 3.0 Pay As You Model. For a small fee, you can then send this encrypted model to us by email for unlocking. Besides the tree representation of your model, this entitles you to receive the code of your model in the programming language of your choice (one of the 8 programming languages available in APS 3.0). This procedure is totally secure and confidential as no information from the datasets used to create the model is transmitted.

To download APS 3.0 Pay As You Model please go to:

http://www.gepsoft.com/PayAsYouModel.htm

Happy Modeling!

2/14/2004 - General features of version 3.0:

• Translates the evolved models into 8 different languages (C, C#, C++, Java, JavaScript, Visual Basic, VB.NET, and Fortran).
• Draws the parse trees of the evolved models.
• Translates the evolved models virtually into any programming language through User Defined Grammars.
• A total of 70 different built-in mathematical functions and comparison rules plus Dynamic UDFs and Static UDFs for modeling.
• A total of 11 built-in fitness functions for Function Finding.
• A total of 10 built-in fitness functions for Classification.
• A total of 11 built-in fitness functions for Times Series Prediction.
• User Defined Fitness Functions for all problem categories.
• Implements a new algorithm for handling random numerical constants.
• Data screening engine for preprocessing.
• Time series transformation engine.
• Evolution from seed models.
• Change seed utilities.
• Saves all the best-of-generation models of a run.
• Plots the evolutionary dynamics of the run.
• Complexity increase engine.
• Supports Databases and Text Files both for loading input data and scoring.
• Recursive testing and prediction for Time Series.
• Implements an extensive package of statistical indexes for model evaluation.
  and much more.
  

You can learn more about APS 3.0 here.

Automatic Problem Solver is a
Microsoft awarded application and
the first commercial software to use
Gene Expression Programming.

7/6/2003 - The book Gene Expression Programming: Mathematical Modeling by an Artificial Intelligence by Candida Ferreira, 2002, has now been published and is available to order online.

In this first book on gene expression programming the author describes thoroughly the basic gene expression algorithm and numerous modifications to this new algorithm, providing all the implementation details so that anyone with elementary programming skills will be able to implement it themselves.

As a powerful meta-language, gene expression programming touches all the fields of computer intelligence and everyone who faces challenging problems and cannot solve them using either traditional mathematical approaches or sophisticated machine learning techniques can benefit from the practical understanding of this new powerful technique.

The book is self-contained and can be used by people with little knowledge of calculus and no prerequisites associated with knowledge of any programming language are required.

The book provides an introduction to this new exciting field of computer intelligence, including a large body of previously unpublished materials such as:

o data mining
o classifier systems
o parameter optimization
o evolution of Kolmogorov-Gabor polynomials
o times series prediction
o evolution of linking functions
o multicellular systems
o automatically defined functions
o user defined functions
o complete neural network induction

The book also discusses some important and controversial evolutionary topics that might be refreshing to both evolutionary computists and evolutionary biologists.

For further information and to order online, please go to:

http://www.gene-expression-programming.com/Books/index.asp

6/9/2003 - Matthew Sottile released into the open source community a Java Gene Expression Programming toolkit. In his words, "My hope is that this toolkit can be used to rapidly build prototype codes that use GEP, which can then be written in a language such as C or Fortran for real speed. I decided to release it as an open source project to hopefully get others interested in contributing code and improving things.". jGEP is hosted at Sourceforge: http://sourceforge.net/projects/jgep/

3/18/2003 - The paper published in Advances in Complex Systems is now available for download at the World Scientific Website

Ferreira, C., 2002. Genetic Representation and Genetic Neutrality in Gene Expression Programming. Advances in Complex Systems, 5 (4): 389-408.

ABSTRACT:
The neutral theory of molecular evolution states that the accumulation of neutral mutations in the genome is fundamental for evolution to occur. The genetic representation of gene expression programming, an artificial genotype/phenotype system, not only allows the existence of non-coding regions in the genome where neutral mutations can accumulate but also allows the controlled manipulation of both the number and the extent of these non-coding regions. Therefore, gene expression programming is an ideal artificial system where the neutral theory of evolution can be tested in order to gain some insights into the workings of artificial evolutionary systems. The results presented in this work show beyond any doubt that the existence of neutral regions in the genome is fundamental for evolution to occur efficiently.

12/9/2002 - The paper presented at the Second International Conference on Hybrid Intelligent Systems (HIS 2002) is now available for download here.

Ferreira, C., Analyzing the Founder Effect in Simulated Evolutionary Processes Using Gene Expression Programming. In A. Abraham, J. Ruiz-del-Solar, and M. Köppen (eds), Soft Computing Systems: Design, Management and Applications, pp. 153-162, IOS Press, Netherlands, 2002.

ABSTRACT:
Gene expression programming is a genotype/phenotype system that evolves computer programs encoded in linear chromosomes of fixed length. The interplay between genotype (chromosomes) and phenotype (expression trees) is made possible by the structural and functional organization of the linear chromosomes. This organization allows the unconstrained operation of important genetic operators such as mutation, transposition, and recombination. Although simple, the genotype/phenotype system of gene expression programming can provide some insights into natural evolutionary processes. In this work the question of the initial diversity in evolving populations of computer programs is addressed by analyzing populations undergoing either mutation or recombination. The results presented here show that populations undergoing mutation recover practically undisturbed from evolutionary bottlenecks whereas populations undergoing recombination alone depend considerably on the size of the founder population and are unable to evolve efficiently if subjected to really tight bottlenecks.

10/7/2002 - The paper presented at the 7th Online World Conference on Soft Computing in Industrial Applications is now available for download at:

http://www.gene-expression-programming.com/author.asp

Ferreira, C., Function Finding and the Creation of Numerical Constants in Gene Expression Programming. 7th Online World Conference on Soft Computing in Industrial Applications, September 23 - October 4, 2002.

ABSTRACT: Gene expression programming is a genotype/phenotype system that evolves computer programs of different sizes and shapes (the phenotype) encoded in linear chromosomes of fixed length (the genotype). The chromosomes are composed of multiple genes, each gene encoding a smaller sub-program. Furthermore, the structural and functional organization of the linear chromosomes allows the unconstrained operation of important genetic operators such as mutation, transposition, and recombination. In this work, three function finding problems, including a high dimensional time series prediction task, are analyzed in an attempt to discuss the question of constant creation in evolutionary computation by comparing two different approaches to the problem of constant creation. The first algorithm involves a facility to manipulate random numerical constants, whereas the second finds the numerical constants on its own or invents new ways of representing them. The results presented here show that evolutionary algorithms perform considerably worse if numerical constants are explicitly used.

9/19/2002 - The paper published in the Proceedings of the Argentine Symposium on Artificial Intelligence is now available for download here.

Ferreira, C., Combinatorial Optimization by Gene Expression Programming: Inversion Revisited. In J. M. Santos and A. Zapico, eds., Proceedings of the Argentine Symposium on Artificial Intelligence, pages 160-174, Santa Fe, Argentina, 2002.

ABSTRACT: Combinatorial optimization problems require combinatorial-specific search operators so that populations of candidate solutions can evolve efficiently. Indeed, several researchers created modifications to the basic genetic operators of mutation and recombination in order to create high performing combinatorial-specific operators. However, it is not known which operators perform better as no systematic comparisons have been done. In this work, a new algorithm that explores a new chromosomal organization based on multigene families is used. This new organization together with several combinatorial-specific search operators, namely, inversion, gene and sequence deletion/insertion, and restricted and generalized permutation, allow the algorithm to perform with high efficiency. The performance of the new algorithm is empirically compared on the 13- and 19-cities tour traveling salesperson problem, showing that the long abandoned inversion operator is by far the most efficient of the combinatorial operators. The efficiency and potentialities of the new algorithm are further demonstrated by solving a simple task assignment problem.

9/15/2002 - A shorter version of the tutorial presented at the 6th Online World Conference on Soft Computing in Industrial Applications is now available for download here.

Ferreira, C., Gene Expression Programming in Problem Solving. In R. Roy, M. Köppen, S. Ovaska, T. Furuhashi, and F. Hoffmann, eds., Soft Computing and Industry - Recent Applications, pages 635-654, Springer-Verlag, 2002.

ABSTRACT: Gene expression programming is a full fledged genotype/phenotype system that evolves computer programs encoded in linear chromosomes of fixed length. The structural organization of the linear chromosomes allows the unconstrained and fruitful (in the sense that no invalid phenotypes will follow) operation of important genetic operators such as mutation, transposition, and recombination as the expression of each gene always results in valid programs. Although simple, the genotype/phenotype system of gene expression programming is the first artificial genotype/phenotype system with a complex and sounding translation mechanism. Indeed, the interplay between genotype (chromosomes) and phenotype (expression trees) is at the core of the tremendous increase in performance observed in gene expression programming. Furthermore, gene expression programming shares with genetic programming the same kind of tree representation and, therefore, with GEP it is possible, for one thing, to retrace easily the steps undertaken by genetic programming and, for another, to explore easily new frontiers opened up by the crossing of the phenotype threshold. In this tutorial, the fundamental differences between gene expression programming and its predecessors, genetic algorithms and genetic programming, are briefly summarized so that the evolutionary advantages of gene expression programming could be better understood. The work proceeds with a detailed description of the main players in this new algorithm, focusing mainly on the interactions between them and how the simple yet revolutionary structure of the chromosomes allows the efficient, unconstrained exploration of the search space.

3/23/2002 - The paper presented at the 4th International Workshop on Frontiers in Evolutionary Algorithms is now available for download here.

TITLE:
Mutation, Transposition, and Recombination: An Analysis of the Evolutionary Dynamics.

ABSTRACT:
Gene expression programming (GEP) uses mutation, transposition, and crossover to create variation. Although there exists a large body of work in genetic algorithms concerning the roles of mutation and recombination, these results not only do not apply to GEP due to the genotype/phenotype representation but also seem to contradict the GEP experience. Therefore, and given the diversity of GEP operators, it is convenient to develop some kind of understanding of their power. The aim of this work is to help develop such an understanding and to show the evolutionary dynamics and the transforming power of each genetic operator, with their advantages and limitations.

3/5/2002 - The new GEPSR 2.0 COM component includes both Classification and Function Finding. This new release allows Visual Basic and C++ code generation, processes 256 independent variables and up to 10,000 experimental values.

For a complete list of the new features go to:
http://www.gepsoft.com/

With GEPSR 2.0 you can easily build powerfull and innovative applications to analyze numerical data with minimal code on your part. The sample application includes code to connect to any ODBC compatible database and the expanded feature set makes GEPSR the most powerfull and versatile AI component in the market.

GEPSR is a COM component developed with the ATL framework. It can be used from Visual Basic and Excel as well as from most other COM compatible languages and tools under Windows 98, Windows ME, Windows NT 4.0 or Windows 2000.

GEPSR 2.0 is available at a reduced price to education institutions.

12/5/2001 - Complete reference for the first GEP paper:
Ferreira, C., (2001). Gene Expression Programming: A New Adaptive Algorithm for Solving Problems, Complex Systems, 13 (2): 87 - 129.

The final version is available for download at:
http://www.gene-expression-programming.com/webpapers/gep.pdf

10/7/2001 - The GEP Online Bibliography is now available to GEP researchers. You can include your GEP publications there.

10/7/2001 - The GEP tutorial presented this last September at the 6th Online World Conference on Soft Computing in Industrial Applications is now available. You can download the paper or the presentation.

10/7/2001 - The Academic Edition of Automatic Problem Solver includes all the features of the Enterprise Edition

You can learn more about APS 2.0 Academic Edition here



Automatic Problem Solver is a
Microsoft awarded application and
the first commercial software to use
Gene Expression Programming.

6/22/2001 - The new Automatic Problem Solver includes Classification and Function Finding and a new edition (Enterprise Edition), which processes up to 256 independent variables. Also, the number of independent variables of the Advanced and Standard editions were raised to 16 and 5, respectively.

Furthermore, it was introduced a new feature to immediately apply the evolved model to an unlimited number of values (Calculation) and now the evolution can be stopped and restarted anytime.

You can learn more about APS 2.0 here



Automatic Problem Solver is a
Microsoft awarded application and
the first commercial software to use
Gene Expression Programming.

5/24/2001 - I am proud to announce that the Automatic Problem Solver won the third prize of the Microsoft Portugal Software Science Award.
APS was released by Gepsoft in March and is an AI tool that uses gene expression programming.

4/16/2001 - A revised and expanded version of the seminal GEP paper was accepted for publication in Complex Systems. You can download this version in pdf format here. C. Ferreira, 2001. Gene Expression Programming: a New Adaptive Algorithm for Solving Problems. Complex Systems, forthcoming.

3/26/2001 - This edition of APS extends the capabilities of the standard APS in the number of terminals available (up to 10) and the number of chromosomes (up to 2048) and adds a facility to weight the functions for each run.
You can see more details here

3/16/2001 - Automatic Problem Solver is a symbolic regression tool that generates Visual Basic, C++ and Karva code from the analysis of any kind of numerical data. You can see more details here.

1/15/2001 - Currently two versions of the component GEP Symbolic Regression are available with more to follow in the next few months. With these components you can easily build powerfull and innovative applications to analyze numerical data with minimal code on your part. And if you just want to apply symbolic regression to your data then the included applications and Excel worksheets allow you to start working right away without programming at all.

1/15/2001 - Executables that show the complete evolutionary history of a run.

11/14/2000 - Download the PDF of the first GEP paper: C. Ferreira, 2000. Gene Expression Programming: a New Adaptive Algorithm for Solving Problems, http://www.gene-expression-programming.com/webpapers/gepfirst.pdf

10/9/2000 - Two new CA rules for the density-classification problem better than any human-written and better than the GP rule. Several space-time diagrams for GEP₁ and GEP₂ rules