US 11,836,601 B2
Systems using hash keys to preserve privacy across multiple tasks
Omar Florez Choque, Oakland, CA (US); and Erik Mueller, Chevy Chase, MD (US)
Assigned to CAPITAL ONE SERVICES, LLC, McLean, VA (US)
Filed by Capital One Services, LLC, McLean, VA (US)
Filed on Jan. 19, 2023, as Appl. No. 18/156,997.
Application 18/156,997 is a continuation of application No. 17/391,538, filed on Aug. 2, 2021, granted, now 11,586,877.
Application 17/391,538 is a continuation of application No. 16/866,268, filed on May 4, 2020, granted, now 11,093,821, issued on Aug. 17, 2021.
Application 16/866,268 is a continuation of application No. 16/404,169, filed on May 6, 2019, granted, now 10,643,122, issued on May 5, 2020.
Prior Publication US 2023/0153578 A1, May 18, 2023
Int. Cl. G06N 3/045 (2023.01); G06N 3/08 (2023.01); G06N 20/00 (2019.01); G06N 3/02 (2006.01); G06N 20/10 (2019.01); G06N 3/084 (2023.01); G06N 3/048 (2023.01)
CPC G06N 3/045 (2023.01) [G06N 3/08 (2013.01); G06N 3/02 (2013.01); G06N 3/048 (2023.01); G06N 3/084 (2013.01); G06N 20/00 (2019.01); G06N 20/10 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A system for using hash keys to preserve privacy across multiple tasks comprising:
one or more processors; and
memory storing instructions that, when executed by the one or more processors, are configured to cause the system to:
provide, to an encoder, a plurality of input observations associated with one or more tasks, each of the plurality of input observations comprising a raw signal and a class label;
generate, via the encoder, a plurality of hash keys associated with the one or more tasks, the plurality of hash keys based on the plurality of input observations;
map, via the encoder, the raw signal to an embedding vector to generate a mapping value;
compare, via the encoder, a similarity of the mapping value to one or more available hash keys of the plurality of hash keys;
determine, based on the comparison, whether the mapping value matches a hash key associated with the class label based on a predetermined confidence threshold;
responsive to determining the mapping value does not match the hash key associated with the class label based on the predetermined confidence threshold, assign the embedding vector as a new hash key to the plurality of hash keys; and
responsive to determining the mapping value matches the hash key associated with the class label based on the predetermined confidence threshold, update one or more hash keys of the plurality of hash keys with an updating function.