|
Artificial Intelligence
The heart of artificial intelligence (AI) research is making intelligent machines - machines that can automate human tasks that require intelligent behavior. Intelligent behavior entails reasoning, planning, problem solving, abstract thinking, language comprehension, and learning. AI research also looks at machines to understand human intelligence. Currently, most AI research at Illinois is focused on practical engineering tasks and includes natural language processing, planning, pattern recognition, and image processing. Computers are good at performing well-defined, repetitive computations but poor at doing complex tasks, like reasoning. Reasoning is the process of drawing conclusions from facts, and the object of automatic reasoning is to write computer programs in such a way that reasoning becomes a computational process. Researchers in machine learning develop computer systems that automatically improve their performance through experience. Robots learn to operate in new environments (e.g., the surface of Mars), systems understand natural language text by reading, and automated discovery systems learn by examining huge databases. Although these are being researched in domain specific areas, the ultimate goal is to produce domain-independent enabling technology for a broad range of computer applications. One area in machine learning that Illinois researchers are involved in is natural language processing - making human speech understandable to a computer. Natural language is what humans use to communicate with each other. So much information is available, in free form text, especially on the Web, but because human languages (like English) are so ambiguous, natural language processing must consider syntax, semantics, and context to determine the meaning of a sentence. Computer vision involves programming a computer to understand a scene or features in an image using pattern recognition, statistical learning, projective geometry, image processing, graph theory, and other fields. One major challenge is to help computers, which "see" in two dimensions, understand the three-dimensional real world. Applications range from medical imaging, visual effects, and robotics, to manufacturing, security and surveillance.
|
|