back to

Seminar: 3/2 - Lotfi Zadeh

11:00 to 12:30 PM      at:  

Department of Electrical Engineering & Computer Science
UC, Berkeley
"A Natural-Language-Based Computational Theory of Perceptions (CTP)"

There is an enormous literature on perceptions. But what is not in existence is a computational theory, that is, a theory in which perceptions play the role of objects of computation. Such a theory, CTP for short, is outlined in this lecture. The point of departure in CTP is the observation that natural language is basically a system for describing perceptions. In CTP, this observation serves as a basis for equating a perception to its description in a natural language. It is this equation that opens the door to construction of a computational theory of perceptions. Perceptions are intrinsically imprecise. Imprecision of perceptions is passed on to natural languages. Existing natural processing techniques cannot deal with semantic imprecision of natural languages. What is needed for this purpose is the concept of Precisiated Natural Language (PNL). The centerpiece of PNL is the concept of a generalized constraint. The calculus of generalized constraints is the core of the computational theory of perceptions. Humans have a remarkable capability to perform a wide variety of physical and mental tasks, e.g., driving a car in city traffic, without any measurement, based solely on perceptions. Mechanization of such tools is one of the principal objectives of the computational theory of perceptions.