Support vector classification with bounded uncertainties in input data
AssigneeSiemens Medical Solution USA, Inc.
InventorJinbo Bi
Life of this patent
Estimating the term…
Abstract
A computer-implemented method for determining a boundary for binary classification includes providing a data set, initializing a value for noise in the data set, and determining a hyperplane dividing the data set and a slack variable given a current value for noise. The method further includes updating the value for noise and the slack variable given the hyperplane, and determining the hyperplane to be the boundary for binary classification of the data set upon determining a termination criterion to be met, wherein elements of the data set are classified according to the boundary.
Claims13
A computer-implemented method for determining a boundary for binary classification comprising:
- providing a data set of vectors corrupted by noise;
- initializing a value for uncertainty for each vector in the data set corrupted by noise;
- determining a hyperplane dividing the data set and a slack variable for each vector in the data set given a current value for uncertainty, wherein a weight of each vector on the hyperplane is varied according to the current value for uncertainty for the respective vector and wherein the weight of a respective vector is reduced with increased uncertainty;
- updating the value for uncertainty and the slack variable for each vector in the data set given the hyperplane;
- and
- determining the hyperplane to be the boundary for binary classification of the data set upon determining a termination criterion to be met, wherein the vectors of the data set are classified according to the boundary.
A computer-implemented method for binary classification comprising determining a linear boundary for binary classification of a data set according to
- min
- w
- ,
- b
- ,
- ξ
- ∑
- i
- =
- 1
- l
-
- ξ
- i
- subject
-
-
- to
- y
- i
-
- (
- w
- T
-
- x
- i
- +
- b
- )
- +
- γ
-
-
- δ
- i
- ≥
- 1
- -
- ξ
- i
- ,
- ξ
- i
- ≥
- 0
- ,
- i
- =
- 1
- ,
- …
-
- ,
- l
- ,
-
- w
-
- 2
- ≤
- γ
- 2
- wherein a plurality of elements of the data set are classified according to the linear boundary, wherein
- l is a predetermined number of training examples;
- (xi, yi) is an observation where an input xi is corrupted with noise and yi is an output;
- (w,b) is the linear boundary where w is a weight vector and b is a bias;
- ξi are slack variables, wherein each slack variable corresponds to a data point of the data set;
- wT x +b is a predictor mapping xi to yi;
- δi is a bound corresponding to uncertainty in the input due to the noise;
- and
- ∥w∥2 ≦γ2 is a quadratic constraint.
A program storage device readable by machine, tangibly embodying a program of instructions executable by the machine to perform method steps for determining a boundary for binary classification, the method steps comprising:
- providing a data set of vectors corrupted by noise;
- initializing a value for uncertainty for each vector in the data set corrupted by noise;
- determining a hyperplane dividing the data set and a slack variable for each vector in the data set given a current value for uncertainty, wherein a weight of each vector on the hyperplane is varied according to the current value for uncertainty for the respective vector and wherein the weight of a respective vector is reduced with increased uncertainty;
- updating the value for uncertainty and the slack variable for each vector in the data set given the hyperplane;
- and
- determining the hyperplane to be the boundary for binary classification of the data set upon determining a termination criterion to be met, wherein the vector of the data set are classified according to the boundary.
Dependencies are inferred from the claim text.