Multi-objective grammar-guided genetic programming for bio-network modelling and evolutionary design

点击翻译按钮获取中文摘要

This research uses artificial intelligence algorithms to automatically construct biological simulations, and solve problems in graphics and design. Genetic programming is an artificial intelligence technology that uses a simulation of Darwinian evolution to create computer programs. The research will improve some of the technical aspects of genetic programming, to improve its effectiveness at solving complex problems that are often very challenging for human beings to accomplish. For example, new ways of representing problem-solving programming languages will be derived. Improvements to evaluating potential solutions to problems will also be invented. One application area of genetic programming that will be studied is the automatic construction of biological networks.For example, given some laboratory data denoting the changing quantities of proteins over time, the goal is to have genetic programming automatically use this data to construct a model that explains how this data arises in the lab. This is a challenging task because such laboratory data is usually very complex and noisy, and difficult to analyze. A goal of this research is to discover improved means for analyzing such data. In addition, new modeling languages will be used by genetic programming, to more effectively construct the models. Applications in evolutionary design will also be investigated. These are applications in which genetic programming can be used to create painterly art, 3D models and structures, and other applications in the fine arts. Such problems share many characteristics of biological modeling. For example, they often are complicated to evaluate, and they can benefit with improved programming language representations within the genetic programming system. The field of evolutionary design has been very popular with my students, and many have proceeded to graduate school to work in the area.