US 12,248,855 B2
Iterative vectoring for constructing data driven machine learning models
Oren Elisha, Hertzeliya (IL); Ami Luttwak, Binyamina (IL); Hila Yehuda, Tel Aviv (IL); Adar Kahana, Natanya (IL); and Maya Bechler-Speicher, Tel Aviv (IL)
Assigned to MICROSOFT TECHNOLOGY LICENSING, LLC, Redmond, WA (US)
Filed by Microsoft Technology Licensing, LLC, Redmond, WA (US)
Filed on Oct. 27, 2022, as Appl. No. 18/050,364.
Application 18/050,364 is a continuation of application No. 16/795,307, filed on Feb. 19, 2020, granted, now 11,514,364.
Prior Publication US 2023/0095553 A1, Mar. 30, 2023
Int. Cl. G06N 20/00 (2019.01); G06Q 30/016 (2023.01)
CPC G06N 20/00 (2019.01) [G06Q 30/016 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A system, comprising:
at least one memory that stores program code; and
at least one processor circuit that executes the program code to:
access an initial vector set, a vector of the initial vector set comprising a first set of features generated by a first feature generating algorithm and a second set of features generated by a second feature generating algorithm;
determine a first aggregated measure of importance for the first feature generating algorithm;
determine a second aggregated measure of importance for the second feature generating algorithm;
based on a determination that the second aggregated measure of importance is higher than the first aggregated measure of importance, generate reallocated features in a reallocated feature set by replacing a feature of the first set of features with a feature generated by the second feature generating algorithm that is not part of the second set of features; and
iteratively output a machine-learning (ML) model trained using vectors that include at least some of the reallocated features in the reallocated feature set until the outputted ML model satisfies an accuracy criteria, wherein the satisfaction of the accuracy criteria is based at least on the vectors that include at least some of the reallocated features in the reallocated feature set.