Answers are grounded in this paper and cite their sources.
By employing a Bayesian approach to the analysis of learning the probability distribution of property vectors, an estimation likelihood computation scheme for the general Gaussian distribution (quadratic adaptive decision surface) is shown optimum. Some results relating the number of learning samples to Type I misclassification errors are included.
No takes yet. Share an insight, caveat, or question.
Daniel G. Keehn (1965) studied this question.
Synapse has enriched 5 related papers. Consider them for comparative context: