Computer Graphics and a Revised Concept of Dependence and Independence
The “information” carried by a computer output and, in particular, by a computer graphic display is quantized. Accordingly, the concepts of functional dependence and stochastic independence have been revised to adjust them to the discrete nature of the “information” that can be received. Some basic “natural” properties of the measure of expected information have been used to establish practical criteria to detect dependence and independence by computer data.
In connection with the size of the “pixels” of a graphic display different levels of dependence can be considered. Practical measures for these levels are suggested.
KeywordsMarginal Distribution Equal Probability Discretized Formulation Uniform Continuity QUANTIZED Formulation
Unable to display preview. Download preview PDF.
- 1.R.M. Haralick, ed., Pictorial data analysis, NATO ASI Series, Series F: Com. and Systems Sciences, No. 4, Springer-Verlag, Berlin, 1983.Google Scholar
- 2.B. Forte and N. Pintacuda, Stochastic independence: from pattern recognition a different characterization, Boll. U.M.I, 1984, 6(3-A), pp. 119 – 123Google Scholar
- 3.B. Forte, M. De Lascurain, and A. K. C. Wong, The best lower bound of the maximum entropy for discretized probability distributions, IEEE Trans. Inform. Theory, to appearGoogle Scholar
- 4.C. Bertoluzza and B. Forte, Mutual Dependence of Random Variables and Maximum Discretized Entropy, unpublished manuscript.Google Scholar
- 5.M. De Lascurain, On the Maximum Entropy Discretization and Its Applications in Pattern Recognition, Ph.D. Dissertation, Dept. of Systems Design, Univ. of Waterloo, Waterloo, Ontario, Canada, 1983Google Scholar
- 6.J. Aczel and Z. Daroczy, On Measures of Information and Their Characterizations, Academic Press, New York, 1975Google Scholar