Back to results
US 2012/0259584 A1Published application
Method for Optimizing Parameters for Detection Systems
AssigneeThe Johns Hopkins University
InventorDavid A. Portnoy
Life of this patent
Estimating the term…
Abstract
A method for generating scatter-plots similar to classic ROC curves to gain an understanding of the tradeoffs between probability of detection and probability of false alarm for detections systems. The method uses a stochastic optimization and an objective function to maximize probabilities of detection for probabilities of false alarm that are arbitrarily close to some target probability of false alarm.
Claims4
1:
- A computer program product, comprising a non-transitory computer usable medium having a computer readable program code embodied therein, the computer readable program code adapted to be executed to implement a method for optimizing a plurality of parameters for a detection system to yield a maximum probability of detection for a plurality of predetermined probabilities of false alarm, the method comprising:
- using a genetic algorithm in combination with a simulated annealing algorithm to minimize an objective function, the genetic algorithm using a chromosome to represent a set of parameter values and the simulated annealing algorithm having a temperature parameter.
2:
- The computer program product as recited in claim 1, wherein the objective function is:
- f
-
- (
- P
- d
- ,
- P
- fat
- ,
- P
- fae
- )
- =
- (
- 1
- -
- m
- )
-
- (
- 1
- -
- P
- d
- )
- +
- m
-
-
- P
- fae
- -
- P
- fat
- P
- fat
-
- where
- m
-
- (
- P
- fat
- ,
- P
- fae
- )
- =
- 1
- -
- exp
-
- (
- -
- (
- P
- fae
- -
- P
- fat
- )
- 2
- 2
-
- σ
- 2
- )
- .
66 more elements
3:
- The computer program product as recited in claim 1, the method further comprising:
- generating a plurality of chromosomes;
- setting an initial state
- setting a temperature for the system;
- calculating a fitness of each of the plurality of chromosomes with respect to each of the plurality of predetermined probabilities of false alarm using the objective function;
- and
- adding each of the plurality of chromosomes to one of the plurality of predetermined probabilities of false alarm for which each of the plurality of chromosomes is most fit.
4 more elements
4:
- The computer program product as recited in claim 3, the method further comprising:
- Performing crossover and mutation using the simulated annealing algorithm on each of the plurality of chromosomes;
- calculating the fitness of each mutated chromosome;
- calculating the fitness of each cross-over and mutated chromosome with respect to each of the plurality of predetermined probabilities of false alarm;
- adding each mutated chromosome to each of the plurality of predetermined probabilities of false alarm for which it is more fit than any of the chromosomes currently in the plurality of predetermined probabilities of false alarm;
- decreasing the temperature of the system for each new chromosome added to a population;
- and
- repeating the above steps until the objective function is minimized.
5 more elements