US 12,229,222 B2
Machine learning classifying of data using decision boundaries
Mukundan Sundararajan, Bangalore (IN); Jignesh K Karia, Thane (IN); Radhika Sharma, Bangalore (IN); and Ravindranath Nemani, Bangalore (IN)
Assigned to INTERNATIONAL BUSINESS MACHINES CORPORATION, Armonk, NY (US)
Filed by INTERNATIONAL BUSINESS MACHINES CORPORATION, Armonk, NY (US)
Filed on Oct. 12, 2021, as Appl. No. 17/499,279.
Prior Publication US 2023/0112298 A1, Apr. 13, 2023
Int. Cl. G06F 18/2411 (2023.01); G06F 18/21 (2023.01); G06F 18/23 (2023.01); G06N 3/04 (2023.01)
CPC G06F 18/2411 (2023.01) [G06F 18/2185 (2023.01); G06F 18/23 (2023.01); G06N 3/04 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method for increasing classifier accuracy in machine learning applications comprising:
training a machine learning (ML) model including a classifier across classes by determining weighted input points for a contributing set to store the position and values for true positive and true negative predications;
receiving input data to the classifier of the machine learning model at runtime;
determining a classification output from the classifier, wherein for the classification output the method compares the values for input during runtime with a sample of inputs stored for training the machine learning model to determine an average distance in spread for the classification output;
determining a class from the classification output having a smallest distance and spread; and
characterizing the class with the smallest distance and spread as a true positive or true negative by comparing the class with the smallest distance and spread with the classification output, wherein if the class having the smallest distance and spread that is smaller the average of the classification output the class is designated a false positive or false negative.