Preservation Project
Ant Framework
This is the project I made to obtain my Licenciatura de Computación (a Bachelor equivalent in Computer Science), from La Facultad de Ciencias in the Universidad Central de Venezuela.
In order to obtain my Computer Science degree (with a minor in Artificial Intelligence) I constructed an Artificial Ants' Colony Framework to solve optimization problems. My work was both on the field of Software Engineer and Artificial Intelligence. Specifically, I worked with the mathematical model derived from the behavior of ants' colonies in search of food. The behavior exhibited by Ants well display an interesting kind of self-organisation model, which can be useful to solve optimization problems, e.g., find the shortest path in a graph. In 2010, my work was selected to be part of the Scientific Conference at my Alma Matter UCV. You can check out the conference website.
Downloads
- Antframework, written in Java, is currently hosted at libai (Artificial Intelligence Library) in sourceforge. You can download it here.
- You can download the entire paper (in Spanish: Trabajo Especial de Grado) here
- You can download the presentation (in Spanish) here
- Finally, you can also check out the GitHub Repository for this project
Framework basado en Colonias de Hormigas artificiales para la resolución de problemas de optimización
Trabajo Especial de Grado para obtener la Licenciatura de Computación, de La Facultad de Ciencias en la Universidad Central de Venezuela. La siguiente es na breve descripción:
Las teorías emergentes sobre el caos explican que, de sistemas gobernados por reglas muy simples, es posible obtener comportamientos y resultados complejos que a simple vista pueden considerarse caóticos y erráticos, sin mucho sentido ni aplicación práctica, pero que al ser estudiados con mayor detenimiento resultan en formas sorprendentes que hacen pensar justo lo contrario. Por ejemplo, en matemáticas, el conjunto de Mandelbrot muestra esta situación: de la iteración y retroalimentación positiva del siguiente polinomio cuadrático complejo: Pc : z → z 2 + c , se obtienen estructuras arduo complejas, que al ser graficadas muestran un mundo sorprendente en sí mismas.
Asimismo, dichas teorías tratan de obtener un conocimiento más preciso sobre sistemas en la naturaleza que a simple vista parecen ser caóticos, pero que funcionan con increíble precisión. En el presente trabajo se tratará con uno de estos sistemas: Las Colonias de Hormigas. Una vez entendida y superada la simplicidad de cada uno de sus miembros y su aparente comportamiento caótico, nos encontramos con que esta sociedad de insectos logra organizarse e interactuar de forma tal de cumplir con los objetivos primordiales de cualquier ser vivo: búsqueda de alimento, organización del trabajo, cuidado de la cría, etc. En particular, se hará una revisión de los modelos computacionales derivados del comportamiento de algunas sociedades de hormigas en la búsqueda de alimento. Sorprendentemente, dichos modelos no sólo sirven como ejemplo del orden dentro del aparente caos, sino que incorporan gran utilidad práctica al momento de resolver diversos problemas de optimización, comunes en el día a día.
En 2010, mi trabajo fue seleccionado como parte de las Jornada Científicas de mi Alma Matter UCV. El siguiente es un enlace al sitio web de la conferencia.
Descargas
- El software Antframework, escrito en Java, se encuentra en libai (Artificial Intelligence Library) en sourceforge. Lo puedes descargar haciendo clic aquí.
- El Trabajo Especial de Grado lo puedes descargar aquí.
- La presentación de la defensa de mi TEG la puedes descargar aquí.
Evolving Ants (EVA)
This is a project I made in the Fall 2011 for the I585 BionInspired Computing course, under the supervision of Dr. Luis Rocha, as part of my Master's Degree in Computer Science at Indiana University, Bloomington. You can also check out the presentation I did for this course.
The main idea was to enhance a plain ACO algorithm (see here and here), by providing each ant with genetic information that it uses together with stigmergic information to construct solutions to a path in a maze. What I wanted to test was whether or not individual memory (genetic information) would enhance or hinder the collective behavior of the colony. It turns out that in average it improves the performance of the algorithm!
If you want to read more about the algorithm, a genetic framework constructed for this project, experiments, etc; you can either download the original paper here EvolvingAntsEnriqueAreyan.pdf or read the online version here.
If you want to test the algorithm yourself or perhaps improve it, see the download section below.
Downloads
- You can download the original paper here: EvolvingAntsEnriqueAreyan.pdf.
- You can download the code here: EVA-dist.zip
- Finally, you can also check out the GitHub Repository for this project.
Once you download the .zip file
GlobalWebTek
GlobalWebTek is a Venezuelan based company born in 2003 whose mission is to help other companies in the creation of effective e-business strategies so that they can potentiate the way they do business online.
I work for this company since its beginnings in 2003 until I decided to create my own company (Esto Es Web) with a similar mission but a different approach. In turn I left my company to pursue further graduate studies in The School of Informatics and Computing, Indiana University.
I worked in several projects while being an employee of GWT (GlobalWebTek). Among the most important are:
- PhotoXS a digital photo handling system based on Netherlands
- Hertz Rent a Car web-based rental system in Venezuela, and
- Net2Client a comprehensive web-based CRM system with WebInABox, a CMS companion.
Esto Es Web
Esto Es Web.com, is a company I founded back in 2008 in Caracas, Venezuela, specialized in the conceptualization, development and marketing of MiWeb, a software for medium and small-sized companies that enables to quickly and easily have a useful presence on the web. I built this software as a result of my previous years of experience in web development, from 2003 to 2008. I wanted my customers to be able to update their corporate or personal websites easily, with the resulting HTML and meta information being as clean and search friendly as possible.
Preservation Project
This project is lead by Dr. Geoffrey Brown on the Computer Science Department at Indiana University, Bloomington. This project is now finished. You can browse a repository here and also here. This work was published at The ninth annual iPres conference on digital preservation hosted by the University of Toronto iSchool (Faculty of Information) in Toronto, Ontario, Canada.
Abstract
Access to many born-digital materials can only be accomplished economically through the use of emulation where contemporaneous software is executed on an emulated machine. For example, many thousands of CD-ROMs have been published containing proprietary software that cannot be reasonably recreated. While emulation is proven technology and is widely used to run both current and obsolete versions of Windows and Unix operating systems, it suffers a fatal flaw as a preservation strategy by requiring future users to be facile with today's operating systems and software.
We have previously advocated assisted emulation as a strategy to alleviate this shortcoming. With assisted emulation, a preserved object is stored along with scripts designed to control a legacy software environment and access to the object enabled through a helper application. In this paper we significantly extend this work by examining, for a large data set, both the cost of creating such scripts and the common problems that these scripts must resolve.