| CPC G06V 10/82 (2022.01) [G06F 16/583 (2019.01); G06F 16/90335 (2019.01); G06F 18/21 (2023.01); G06N 3/045 (2023.01); G06N 3/08 (2013.01); G06V 10/454 (2022.01); G06V 10/764 (2022.01); G06V 40/16 (2022.01)] | 20 Claims |
|
1. A method in which one or more processing devices of an authentication server system perform operations comprising:
servicing an authentication query by matching an unstructured input data object included as a query parameter in the authentication query to an unstructured reference data object stored in a memory device of the authentication server system, wherein matching the unstructured input data object to the unstructured reference data object comprises:
providing the unstructured input data object to a dual deep learning network, wherein the dual deep learning network is configured to determine a similarity between unstructured data objects based on embedded features in the unstructured input data object and features of the unstructured reference data object,
applying a relationship subnetwork of the dual deep learning network to an input feature vector and a reference feature vector, wherein the input feature vector and the reference feature vector are generated by an embedding subnetwork of the dual deep learning network, and
obtaining, from the dual deep learning network, an output probability of the unstructured input data object and the unstructured reference data object belonging to a common class; and
transmitting, to a client computing system, a responsive message to the authentication query that is generated from the output probability, wherein the responsive message is configured for causing the client computing system to grant or deny access to the client computing system by a user device from which the unstructured input data object was obtained.
|