Calculating posterior probability of classes
AssigneeLongsand Limited
InventorsGeorge SAKLATVALA, Pashmina CAMERON
Life of this patent
Estimating the term…
Abstract
A plurality of binary support vector machines (SVM) are trained to vote between two of a plurality of classes. Each of the binary SVMs vote between a different pair of the plurality of classes. A set of validated samples is input to each of the binary SVMs. The set of validated samples includes samples of each of the classes. Likelihood estimates are determined based on votes of the binary SVMs. A prior probability distribution (prior) is assigned for each of the classes. A posterior probability is calculated for each of the classes based on the determined likelihood estimates and the assigned priors.
Claims15
A method, comprising:
- training a plurality of binary support vector machines (SVM) to vote between two of a plurality of classes, each of the binary SVMs to vote between a different pair of the plurality of classes;
- inputting a set of validated samples to each of the binary SVMs, the set of validated samples to include samples of each of the classes;
- determining likelihood estimates based on votes of the binary SVMs in response to the inputting;
- assigning a prior probability distribution (prior) for each of the classes;
- and
- calculating a posterior probability for each of the classes based on the determined likelihood estimates and the assigned priors.
A system, comprising:
- a plurality of binary support vector machines (SVM) units to vote between two of a plurality of classes, each of the binary SVMs to vote between a different pair of the plurality of classes;
- a validated sample unit to input a set of validated samples to each of the binary SVMs, the set of validated samples to include samples of each of the classes;
- a posterior probability unit to calculate a posterior probability for each of the classes using Bayesian inference and based on likelihood estimates determined from votes of the binary SVMs in response to the set of validated samples.
A non-transitory computer-readable storage medium storing instructions that, if executed by a processor of a device, cause the processor to:
- run a set of validated samples against each of a plurality binary SVMs, each of the binary SVMs to vote between a different pair of classes of a plurality of classes in response to the input set of validated samples;
- determine likelihood estimates based on the votes of the binary SVMs;
- assign a prior probability distribution (prior) for each of the classes;
- and
- input the determined likelihood estimates and the assigned prior probabilities in to a Bayesian inference model to calculate a posterior probability for each of the classes.
Dependencies are inferred from the claim text.