For more details, check out GECCO-2009 web site. Download the PDF of the GECCO-2009 Call for Papers here. The photo for the GECCO-2009 banner is by Diliff. The original is available from wikipedia.

Básicamente, se ha partido de un controlador heurístico (a la izquierda) el cual se ha mejorado y optimizado mediante técnicas de optimización basadas en colonias de hormigas.

La mejora (derecha) es bastante notable, pero todavía queda margen para arañar unas decimillas...

Luis

April 15-17, 2009, Tübingen, Germany
Homepage: www.evostar.org
Deadline November 5, 2008

FOGA X - Foundations of Genetic Algorithms

January 9-11, 2009, Orlando, Florida USA
Homepage: http://www.sigevo.org/foga-2009
Deadline August 18, 2008

2009 IEEE Congress on Evolutionary Computation (CEC 2009)

May 18-21, 2009, Trondheim, NORWAY
Homepage: www.cec-2009.org
Deadline November 1, 2008

New Journal: IEEE Transactions On Computational Intelligence And AI In Games

Homepage: http://ieee-cis.org/pubs/tciaig/ (available soon)
Submissions open August/September 2008

Luis

come to think of it, sifat ini mirip dengan yang saya alami di blogosphere. Dalam kondisi ideal jika sekumpulan blogger melakukan eksplorasi di bidang yang sama dengan daerah yang dieksplor berbeda, hal ini dapat mempercepat proses belajar (tentunya dengan asumsi tidak ada yang berbohong).

Mungkin ini ya yang dipercaya oleh para peneliti di bidang Multi-Agent Intelligence. Bahwa interaksi antar agent dapat mempercepat konvergensi suatu proses pembelajaran. Adakah yang memiliki paper tentang hubungan ini (fusion of learning experience to speed up convergence in learning multi-agent system)?

saya sendiri belum cukup banyak menjelajahi daerah multi-agent system dalam bidang AI. Tapi kemarin sempat baca-baca sekilas tentang monte-carlo simulation dan markov random field. Sepertinya ada kemiripan dengan proses fusi pembelajaran yang barusan.

*bongkar-bongkar koleksi pustaka*

NB: kok ujung-ujungnya malah ke soft-computing yah? hehe.. biarin ah!

Uno de ellos, Shumeet Baluja es conocido por haber propuesto el  PBIL, uno de los Algoritmos de Estimación de Distribuciones de referencia.

El artículo que lo describe puede consultarse aquí.

10th International Conference on Parallel Problem Solving From Nature
(PPSN 2008)
September 13-17, 2008
Technische Universität Dortmund, Germany
Submission due: April 14, 2008
Conference Website

The Workshop on AI in Games AIG-08
in conjunction with the 18th European Conference on Artificial Intelligence (ECAI-08)
Patras, Greece
July, 2008
Submission due: May 1st, 2008
Conference Website

1st International Workshop on ombinations of Intelligent Methods and Applications
(CIMA 2008)
in conjunction with the 8th European Conference on Artificial Intelligence (ECAI-08)
July 21/22, 2008, Patras, Greece
Submission Due: May 5th, 2008
Conference Website
The 20th IEEE INTERNATIONAL CONFERENCE on TOOLS with ARTIFICIAL INTELLIGENCE
(ICTAI 2008)
November 3-5, 2008
Dayton, Ohio, USA
Submission due: June 2, 2008
Conference Website

IEEE Symposium on Computational Intelligence and Games
(CIG'08)
15-18 December 2008
Perth, Australia
Submission due: 15 July 2008
Conference Website

El resto de los workshops del ECAI pueden ser consultados aquí: http://www.ece.upatras.gr/ecai2008/acworkshops.htm

Dicevo: premesso questo, credo che dovrete sopportarmi ancora per un bel po'. O almeno spero.

• Juegos
• Navegación
• Minería de Datos
• Optimización

El WCCI (World Congress on Computational Intelligence) se celebra este año en Hong Kong. Este congreso reune cada dos años tres conferencias distintas: IEEE Congress on Evolutionary Computation, el IEEE International Conference on Fuzzy Systems y el IEEE International Joint Conference on Neural Networks.

Swarm intelligence (SI) is an artificial intelligence technique based around the study of collective behavior in decentralized, self-organized systems.

SI systems are typically made up of a population of simple agents interacting locally with one another and with their environment. Although there is normally no centralized control structure dictating how individual agents should behave, local interactions between such agents often lead to the emergence of global behavior. Examples of systems like this can be found in nature, including ant colonies, bird flocking, bacterial growth, and fish schooling.

(source: Wikipedia)

Several SI-based algorithms are proposed, like Ant Colony Optimization (ACO), Particle Swarm Intelligence (PSO), Stochastic Diffusion Search (SDS), etc. These algorithms are applied to an impressive number of optimization problems.

ACO is based on the behavior of real ants for finding food. This behavior can be described as follow (a little long, but intresting :D ):

In the real world, ants (initially) wander randomly, and upon finding food return to their colony while laying down pheromone trails. If other ants find such a path, they are likely not to keep traveling at random, but to instead follow the trail, returning and reinforcing it if they eventually find food.

Over time, however, the pheromone trail starts to evaporate, thus reducing its attractive strength. The more time it takes for an ant to travel down the path and back again, the more time the pheromones have to evaporate. A short path, by comparison, gets marched over faster, and thus the pheromone density remains high as it is laid on the path as fast as it can evaporate. Pheromone evaporation has also the advantage of avoiding the convergence to a locally optimal solution. If there were no evaporation at all, the paths chosen by the first ants would tend to be excessively attractive to the following ones. In that case, the exploration of the solution space would be constrained.

Thus, when one ant finds a good (short, in other words) path from the colony to a food source, other ants are more likely to follow that path, and positive feedback eventually leaves all the ants following a single path. The idea of the ant colony algorithm is to mimic this behavior with "simulated ants" walking around the graph representing the problem to solve.

(see a good example here.)

You can find my report about Swarm Intelligence here. It introduces SI in general, and Ant Colony Optimization in detail. Also, an ACO-based for Traveling Salesperson problem and a ACO-based routing algorithm, are described.

More information:
- M. Dorigo & G. D. Caro's paper, the original paper on ACO
- M. Dorigo home page

In classical set theory, an element either belongs or does not belong to the set. By contrast, fuzzy set theory an element belongs to the set, with membership degree in interval [0-1].

Richard, 20
frank, 18
Anna, 14
Juli, 11

We want to define "A = set of Good Student in Biology Class":
1- in classical set theory: A = {Richard, Frank}
2- in fuzzy set theory: A = { (Richard,1) , (Mark,0.7) , (Anna,0.3) , (Juli,0.1) }

In fact, we have associated a membership degree to each member, according to goodness of their score.

Fuzzy logic is derived from fuzzy set theory dealing with reasoning that is approximate rather than precisely deduced from classical predicate logic.

Fuzzy concept has application in many areas, specially in control systems.
In many topics, after proposing an algorithm, the fuzzy version of the algorithm is proposed and evaluated against the original one.

Here, is our report, in Persian, that introduces data mining in general, Apriori algorithm to extract association rules, a fuzzy-based Apriori algorithm, Decision Tree and fuzzy-based Decision Tree.

More information:
- Zadeh's 1965 paper on Fuzzy Sets (It has more that 5000 citations till now!)

Questo solo per (mia) comodità e per sopperire alla mancanza di statistiche sui feeds da parte di Wordpress.

Ringrazio tutti per la pazienza e vi saluto perchè devo andare a studiare "Soft Computing".

P.S.: Naturalmente pausa studio programmata per le 18:00 in concomitanza con il posticipo del Milan.

If you trying to build Computer Vision apps, these libraries will shorten your project time :

• OpenCV (Open Source Computer Vision Library). This libray covers large areas of computer vision. Some demos (apps that come with the library) are included. The only reason why i use this library for my face recognition project is the community where i can get answer from annoying problems. OpenCV Wiki is a good start to get answer, then you can go to OpenCV Forums at Yahoo Groups.
• Torch3vision (The Vision Package of Torch), a machine vision library based on Torch (a machine learning library written in simple C++ and developed at IDIAP). Sebastien Marcel, one of the developers, gives some demos based on real-time face detection & recognition using Torch3vision available on his homepage. It's awesome dude. He also gives information in comparison between OpenCV face detector and IDIAP face detector.
• MPT (Machine Perception Toolbox) which is developed at machine perception laboratory of UCSD. It supplies cross-platform libraries for real-time perception primitives, including face detection, eye detection, blink detection, color tracking. Soon it will also include expression recognition, predictive color tracking, and tracking based on multisensor fusion. You should see some examples (screenshoots of application) which use this library.
• Collection of computer vision software (with source code) at School of Computer Science / CMU.

So why you should build code which already built by top guys? If you have suggestions to be listed above just tell me, or you maybe interested to join with me in my face recognition project :D

• Contoh 1. Gambar ditengah merepresentasikan membership function yang dibangun pada kurva distribusi Gaussian. Disebelah kanan dan kiri terdapat jaringan neural sederhana.
  File psdnya
• Contoh 2. Yang saya buat mencolok hanya bentuk triangular membership function
  File psdnya

Maaf hanya bisa dua dulu, jika ada masukan silahkan tinggalkan pesan :D.

Genetic Algorithm (GA) is a search technique to find true or approximate solution, inspired by evolutionary biology such as inheritance, mutation and recombination.

The concept is very simple. Assume we have a problem and a pool of solutions (search space). Typically solution are represented by 1s and 0s.
Each individual has a fitness, according to a defined criteria.
GA does many iterations and terminates when reaches a satisfactory fitness level.

Definitions:
mutate ==> changing 1 to 0 and vise versa in a bit string
crossover ==> combining 2 string (parents) and generating
2 new ones (offspring)

The steps in summary are:
while (max fitness of current generation is below of desired fitness) {
1- select probabilistically some individuals from current population
2- recombine them 2 by 2 (crossover)
3- mutate some of them
4- replace new generation with old one
}

You can see a graphical representation of GA in GA viewer.

An implementation of GA in Java, for playTennis problem: Java Class - Example file for testing

Note: It's not well-documented, I wrote it quickly!