Knowledge-Based Intelligent Information and Engineering Systems

Volume 3683 of the series Lecture Notes in Computer Science pp 1143-1150

Ranking Functions, Perceptrons, and Associated Probabilities

  • Bernd-Jürgen FalkowskiAffiliated withUniversity of Applied Sciences Stralsund

* Final gross prices may vary according to local VAT.

Get Access


The paper is motivated by a ranking problem arising e.g. in financial institutions. This ranking problem is reduced to a system of inequalities that may be solved by applying the perceptron learning theorem. Under certain additional assumptions the associated probabilities are derived by exploiting Bayes’ Theorem. It is shown that from these a posteriori probabilities the original classifier may be recovered. On the other hand, assuming that perfect classification is possible, a maximum likelihood solution is derived from the classifier. Some experimental results are given.