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
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-11755709-B2
Publication Date
2023-09-12
Expiration Date
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 relates to establishing authentication credentials using a plurality of non-deterministic registration biometric inputs. During registration, a pre-trained machine learning model usable for a plurality of users generates sets of feature vectors from the non-deterministic biometric inputs, where the inputs include a plurality of face images and a plurality of voice samples of a user. The sets of feature vectors are projected onto a surface of a hyper-sphere.
A characteristic identity vector representing the user is computed based on the user's set of the projected feature vectors. The characteristic identity vector is saved for use during authentication of the user, so that later authentication is performed by comparing authentication embeddings derived from newly provided biometric inputs to the stored characteristic identity vector.
The document further supports registering cryptographic signatures of an identity vector with an identity server, including associating the identity with a user photograph selected from multiple face images. It also describes preprocessing for face and voice biometrics, including face/background separation, selecting sharp video frames based on an image sharpness metric, and selecting a voice audio segment that matches a user-specified phrase to be read. Alternative matching using distance-preserving locality-preserving hashing and/or binning plus hash quantization is also described, producing deterministic registration/authentication hashes for matching.
Claims Coverage
The document includes three independent claims. Each independent claim centers on the same authentication-credential pipeline based on non-deterministic biometric inputs, feature vectors from a pre-trained machine learning model, hyper-sphere projection, computing a characteristic identity vector, and saving it for authentication.
Non-deterministic biometric registration feature vectors into hyper-sphere projection and characteristic identity vector storage
During registration, the plurality of non-deterministic biometric inputs are fed to a pre-trained machine learning model usable for a plurality of users, generating sets of feature vectors. The non-deterministic biometric inputs include a plurality of face images and a plurality of voice samples of a user. The sets of feature vectors are projected onto a surface of a hyper-sphere, a characteristic identity vector representing the user is computed based on a user's set of the projected feature vectors, and the characteristic identity vector is saved for use during authentication of the user.
System instructions implementing non-deterministic biometric registration hyper-sphere projection and characteristic identity vector storage
A system includes one or more processors coupled to memory, and the memory is loaded with computer instructions to establish authentication credentials using a plurality of non-deterministic registration biometric inputs. The instructions implement actions comprising feeding the plurality of non-deterministic biometric inputs to a pre-trained machine learning model usable for a plurality of users, generating sets of feature vectors from a plurality of face images and a plurality of voice samples of a user, projecting the sets of feature vectors onto a surface of a hyper-sphere, computing a characteristic identity vector representing the user, and saving the characteristic identity vector for use during authentication of the user.
Non-transitory storage medium instructions for non-deterministic biometric registration hyper-sphere projection and characteristic identity vector storage
A non-transitory computer readable storage medium is impressed with computer program instructions to establish authentication credentials using a plurality of non-deterministic registration biometric inputs. When executed on a processor, the instructions implement a method comprising feeding the non-deterministic biometric inputs to a pre-trained machine learning model usable for a plurality of users, generating sets of feature vectors from a plurality of face images and a plurality of voice samples of a user, projecting the sets of feature vectors onto a surface of a hyper-sphere, computing a characteristic identity vector representing the user, and saving the characteristic identity vector for use during authentication of the user.
Across the independent claims, the inventive approach is the generation of feature vectors from non-deterministic registration biometrics, hyper-sphere projection, computation of a characteristic identity vector, and storage of that vector for later authentication.
Stated Advantages
Documented Applications
No documented applications found
Interested in licensing this patent?