Estimating the Support of a High-Dimensional Distribution

  • John Platt ,
  • Bernhard Schölkopf ,
  • John Shawe-Taylor ,
  • Alex J. Smola ,
  • Robert C. Williamson

MSR-TR-99-87 |

Suppose you are given some dataset drawn from an underlying probability distribution P and you want to estimate a “simple” subset S of input space such that the probability that a test point drawn from P lies outside of S is bounded by some a priori specified v between 0 and 1. We propose a method to approach this problem by trying to estimate a function f which is positive on S and negative on the complement. The functional form of f is given by a kernel expansion in terms of a potentially small subset of the training data; it is regularized by controlling the length of the weight vector in an associated feature space. The expansion coefficients are found by solving a quadratic programming problem, which we do by carrying out sequential optimization over pairs of input patterns. We also provide a preliminary theoretical analysis of the statistical performance of our algorithm. The algorithm is a natural extension of the support vector algorithm to the case of unlabelled data.