Signal-to-Symbol Transformation Processes to Representation RUZENA BAJCSY Department ( bajcsy~ of Computer central. and In formation Science, and Vice Versa: From Fundamental UnlversLty of Pennsyluanza, Phlladelphla, PA 19104 czs.upenn,edu) Introduction It is a trivial but true statement that although the physical world in which we and artificial agents exist is continuous and is perceived or sensed as a continuous signal, our reasoning is performed on symbolic entities. All natural and artificial sensors first detect continuous signals. Classical physics, chemistry, and biology describe some processes and behaviors, at least at a certain level of granularity, using continuous mathematics. Symbolic information processing, on the other hand, utilizes discrete mathematics. One way of looking at the problem of signal-to-symbol transformation is as a problem of quantization. Hence, we believe that to understand knowledge representation fully we must fh-st underto stand this transformation from signal symbol and vice versa. Problem Definition a purely random process. Finally, we must answer the question of how fixed or fluid this partitioning is, and thus the symbolic alphabet. That is, how much is task- and context-dependent? The Perception-Action Cycle The first problem is determining the symbolic alphabet into which we partition or segment the signal. (Those
/lp/association-for-computing-machinery/signal-to-symbol-transformation-and-vice-versa-from-fundamental-AAEOGOe8SF