Artificial intelligence-based generation of anthropomorphic signatures and use thereof

Inventors

SLY, Axel • SHARMA, Srivatsa Akshay • REDINGER, Brett Robert • REICH, Devin Daniel • Trooskens, Geert • LOOTUS, Meelis • Lee, Young Jin • ARREDONDO, Ricardo Lopez • KAUTZ, IV, Frederick Franklin • BHAT, Satish Srinivasan • KIRK, Scott Michael • DE BROUWER, Walter Adolf • THAKORE, Kartik

Assignees

DocAI Inc • Sharecare AI Inc

Abstract

The technology disclosed relates to authenticating users using a plurality of non-deterministic registration biometric inputs. During registration, a plurality of non-deterministic biometric inputs are given as input to a trained machine learning model to generate sets of feature vectors. The non-deterministic biometric inputs can include a plurality of face images and a plurality of voice samples of a user. A characteristic identity vector for the user can be determined by averaging feature vectors. During authentication, a plurality of non-deterministic biometric inputs are given as input to a trained machine learning model to generate a set of authentication feature vectors. The sets of feature vectors are projected onto a surface of a hyper-sphere. The system can authenticate the user when a cosine distance between the authentication feature vector and a characteristic identity vector for the user is less than a pre-determined threshold.

Core Innovation

The invention establishes authentication credentials using a plurality of non-deterministic registration biometric inputs and at least one deterministic biometric input. During registration, a trained machine learning model receives sets of feature vectors generated from a plurality of face images and a plurality of voice samples, together with deterministic genomic data of a user. The approach generates projected feature representations by handling the non-deterministic biometric inputs alongside the deterministic genomic data through the same trained machine learning model.

The invention projects the sets of feature vectors onto a surface of a hyper-sphere and computes a characteristic identity vector representing the user based on a user's set of the projected feature vectors. The characteristic identity vector is then saved for use during authentication of the user. The projection is specified in dependent forms as onto a surface of a unit hyper-sphere, and genomic data can be incorporated by feeding genomic data into a hash and using hashed genomic data as at least one dimension of the characteristic identity vector.

The invention secures the workflow by using cryptographic signing of the identity representation and storing encrypted local data, and by generating and validating time-limited scannable codes using hashes with server-side verification to mitigate replay and credential reuse. The document also supports optional presentation of an identity photograph for human verification. During authentication, authentication feature vectors are compared to the stored characteristic identity vector, including via cosine distance, and alternative comparison mechanisms are described using distance-preserving hashing/locality-preserving hashing and binning with hashing with deterministic dimensions.

Claims Coverage

The independent claims cover a computer-implemented registration and enrollment method, a corresponding system, and a non-transitory computer readable storage medium. The core coverage includes multiple non-deterministic biometric inputs and at least one deterministic biometric input, generation of feature vectors using a trained machine learning model, hyper-sphere projection, computation of a characteristic identity vector, and saving that identity vector for authentication.

The independent claims share the same credential-establishment pipeline: use non-deterministic face images and voice samples together with deterministic genomic data, generate feature vectors using a trained machine learning model, project the feature vectors onto a hyper-sphere, compute a characteristic identity vector, and save it for authentication. Dependent claim coverage further tightens the projection geometry, specifies hashing-based incorporation of genomic data, and adds registration features such as cryptographic signature registration with an identity server and associating a user photograph.

Stated Advantages

Documented Applications

No documented applications found