| CPC G06N 20/20 (2019.01) [G06F 18/217 (2023.01); G06F 18/2433 (2023.01); G06F 18/285 (2023.01); G06N 5/022 (2013.01); G06N 20/00 (2019.01); H04L 63/102 (2013.01); H04L 63/12 (2013.01)] | 20 Claims |
|
1. A method that includes one or more processing devices performing operations, the method comprising:
receiving, by a unit-classification system, an identity component (“IC”) data set that includes multiple identity data objects, each one of the multiple identity data objects corresponding to a personal identity component;
determining, by the unit-classification system, multiple feature vectors of the multiple identity data objects, wherein a respective feature vector for a respective identity data object indicates extracted features of the respective identity data object, the extracted features further including a number pattern feature indicating one or more shared characteristics of a particular group of identity data objects;
training a machine learning model based on the multiple feature vectors, wherein the training includes:
generating a waterfall-structure classifier model including a first one-class classifier and a second one-class classifier, wherein the waterfall-structure classifier model is configured to determine, based on each feature vector for each respective identity data object, that the respective identity data object is included in a respective IC category that identifies an inclusion or an exclusion of the respective identity data object in a multi-dwelling unit;
generating, by the unit-classification system, an IC identification of a first multi-dwelling unit that is included in a first IC category indicated by the first one-class classifier, wherein the IC identification is associated with a first identity data object corresponding to a first personal identity component, and wherein the first IC category represents a group comprising multiple personal identity components;
identifying, by the unit-classification system and based on a second personal identity component being included in the first IC category, a second identity data object that corresponds to (i) the second personal identity component and (ii) the IC identification of the first multi-dwelling unit;
modifying, by the unit-classification system, the first identity data object and the second identity data object to include the IC identification of the first multi-dwelling unit; and
providing, to a requesting computing system and responsive to a request for the IC identification of the first multi-dwelling unit, a response indicating the modified first identity data object and the modified second identity data object, wherein the requesting computing system is configured for verifying an electronic persona based on the response, the verifying including:
determining that the first IC category in which the second identity data object is included is atypical of the first multi-dwelling unit based on a comparison of a feature of the second identity data object with a feature of the first identity data object, wherein the feature comprises a latitude/longitude feature, and wherein the first data identity object and the second identity data object are associated with the IC identification of the first multi-dwelling unit,
determining that a particular electronic persona including the second personal identity component is synthetic based on the determination that the first IC category is atypical of the first multi-dwelling unit, and
restricting access, by a target entity corresponding to the particular electronic persona, to a function of the requesting computing system.
|