Abstract: Particular embodiments receive, by a payment service in association with a user account associated with a service provider, an identity verification submission associated with an identifier and a name. The payment service, for the phone number, queries contact records associated with user accounts associated with the service provider. The payment service receives, as a result of the query, several name options associated with the phone number, where each of the name options are stored in individual contact records. The payment service determines a name option of the several name options to associate with the identifier based at least in part on a similarity of individual name options to other name options. The payment service approves or rejects the identity verification submission based at least in part on the determined name option.

Abstract: Using face grouping for fraud reduction is described. A server(s) may receive, in association with a request to access the service, image data representing a face of a user and additional data, determine, based at least in part on the image data and using a trained machine learning model(s), a representation of the face, and determine one or more representations of faces associated with the representation. The server(s) can further determine first information based at least in part on the additional data, determine second information associated with the one or more representations, determine one or more differences between the first information and the second information, and determine whether to accept or deny the request based at least in part on the determining of the one or more differences.

Abstract: Systems and methods are disclosed for content analysis and/or processing. In some examples, a system receive content (e.g., an image) that includes a representation of an object (e.g., a person) in a scene. The system analyzes the content to detect a first portion of the content (e.g., subset of pixels) that represents the object. A second portion of the content represents the scene. The system obfuscates (e.g., masks, blurs, pixelates, replaces) the first portion of the content to generate obfuscated content. The obfuscated content retains the second portion of the content that represents the scene. The system analyzes the obfuscated content using at least one trained machine learning model to generate a prediction of a geographic region (e.g., at least a portion of a continent) that the scene is in. The system outputs an indicator of the prediction of the geographic region.