Expected Long Term Benefits of the Proposal



Expected Long Term Benefits of the Proposal

Evolutionary computing has been used in engineering, particularly in optimization, to solve computationally hard problems. With experience, genetic algorithms can be applied as a general-purpose method across disciplines. Genetic algorithms are also called “adaptive” algorithms and are therefore part what is generally referred to as “adaptive methods” in computer science. Artificial Neural Networks (ANNs) are another area in adaptive methods, which has received attention over the last ten years, or so. The underlying approach involves looking at biological systems to gain insights for the creation of computer systems, and more specifically to introduce elements of randomization and adaptation into the computational process.

I have a strong background in linear programming and optimization. (I used to work on the optimization of airline crew scheduling at American Airlines.) Since simulated annealing and tabu search - optimization techniques that introduce randomization into the optimization process - are part of traditional optimization, it is somewhat natural that I developed an interest in the area of adaptive algorithms.

Work in the area of genetics and adaptive algorithms will provide a strong complement to my expertise in theoretical areas of computer science. In order to move into the field I have undertaken the following steps since I came to UNLV:

1. I have attended a workshop at UCLA Extension, which introduced me to GALib, a C++ library of genetic algorithm objects, which was recently developed at MIT. Dean Wells funded me for this.

2. I have developed the course CSC 789.001 “Genetic Algorithms and Neural Networks,” which I am currently teaching. The course is described further at .

3. I was also able to interest a student (Bradley Hendricks) and he has started work on a masters project “Genetic Algorithms using GALib” which I am now supervising. With him I am setting up the software environment to use on future projects.

4. In October (only a few weeks after I had joined UNLV as a faculty member,) I had submitted a proposal to the Honda Initiation Grant program under the title “Adaptive Algorithms for Automotive Design.” Unfortunately, Honda funded only 6% of the submitted proposals.

To get this effort off the ground funding through the NIA program is absolutely essential. The initiative will make it possible to create a solid foundation for a long term research program in adaptive algorithms. This is my vision for this work

1. Unlike fundamental work in theoretical computer science, my original area, genetic algorithms can be readily used across disciplines, especially disciplines in engineering. Much work related to genetic algorithms has been pursued in Mechanical and Civil Engineering. In Electrical Engineering genetics algorithms can be used in the studey of wavelets.

2. I see an increasing number of Masters students involved in this work, precisely because of the applied nature of the work, and the opportunity for students to do “hands on” design and implementation.

3. The work proposed here will make it possible to continue work in many directions and broaden the scope of my approach.

I expect, aided by the SITE funds, to, especially develop and create software, and to submit some of the ideas from that proposal to the national funding agencies. What is needed now is the acquisition of various software packages, as well as literature, especially technical manuals, and limited funding to further involve a graduate student into the programming.

Such software creation and preliminary results from the SITE project will make a convincing case for funding to these agencies.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download