Mark R. Titchener

Dr. Mark R. Titchener

Centre for Image Technology and Robotics

Department of Computer Science

University of Auckland, New Zealand

Thursday, April 13
4:00 - 5:00 P.M.
Room 1001 EECS

Towards a scientific theory of information

Modern information theory is founded principally on the work of Shannon, who provided a probabilistic framework for information, justified largely in terms of the reasonableness of its implications. Nevertheless, probabilities are not physical quantities that can be measured, like mass, distance, or energy, but a mathematical notion applied in the modeling of complex physical systems in such a way as to abstract out the underlying mechanisms and processes. However, Cherry provides an interpretation of Shannon's entropy which allows us to move out of a probabilistic framework into a counting domain. In this talk, I introduce a recursive hierarchical pattern copying (RHPC) algorithm, and infer from experimental results a direct relationship between the number of RHPC steps required to construct a string and its information content. The result is a theory directly applicable to individual strings and useful in the practical investigation of information structures and processes.

return to CSPL Seminars homepage