Aplicação de inteligência artificial em modelagem científica

Abstract

Artificial Intelligence (A.I.) is a growing field of research aiming to make intelligent agents which can solve hard problems and auxiliate in computational activity. Neural network is an A.I. field inspired by concepts of intelligence in biological neurons, simulating the brain behavior. The main feature of neural networks is their ability to learn and adapt itself to a problem of interest. This field have accomplished remarkable success in the past few years, like visual computation and game theory. Despite this growing number of applications in computer science, A.I. applied to scientific is just beginning and remains unknown to most people. This work considers A.I. applications in scientific modeling. Function approximation and inversed pendulum were treated as a simple model to study and weather forecasting as a hard problem of main interest. The results were analyzed with respect to the type of neural network and their topology, like the number of neurons. We have shown that artificial intelligence is abble to solve certain kinds of problems, but there are restrictions of how far it can go. Lastly, the conclusion discuss the quality of the results and points out how the limitations of A.I. can be circumvented.