US 10,891,521 B2
Adjusting training set combination based on classification accuracy
Fardin Abdi Taghi Abad, Champaign, IL (US); Jeremy Edward Goodsitt, Champaign, IL (US); and Austin Grant Walters, Savoy, IL (US)
Assigned to Capital One Services, LLC, McLean, VA (US)
Filed by Capital One Services, LLC, McLean, VA (US)
Filed on Dec. 6, 2019, as Appl. No. 16/705,222.
Application 16/705,222 is a continuation of application No. 16/441,649, filed on Jun. 14, 2019, granted, now 10,534,984.
Application 16/441,649 is a continuation of application No. 16/151,825, filed on Oct. 4, 2018, granted, now 10,402,691, issued on Sep. 3, 2019.
Prior Publication US 2020/0110969 A1, Apr. 9, 2020
Int. Cl. G06K 9/00 (2006.01); G06K 9/62 (2006.01); G06N 20/00 (2019.01)
CPC G06K 9/6262 (2013.01) [G06K 9/628 (2013.01); G06K 9/6253 (2013.01); G06K 9/6257 (2013.01); G06N 20/00 (2019.01)] 20 Claims
OG exemplary drawing
 
1. An apparatus, comprising:
processing circuitry operable to execute stored instructions that, when executed, causes the processing circuitry to:
perform a first training of a classification model using a training batch, the training batch including a plurality of samples associated with one or more classes of the classification model;
determine whether each class of the classification model meets or exceeds a predefined accuracy level based at least in part on the first training;
determine, dynamically, a number of additional samples to allocate to each class that does not meet the predefined accuracy level based at least in part on a proportion-based analysis; and
adjust the training batch such that the determined number of additional samples are included in the training batch to provide an adjusted training batch.