Visa International Service Association

Peter Rindal

Pratyay Mukherjee

Saikrishna Badrinarayanan

Daniel Siegfried Werner Masny

Shashank Agrawal

Srinivasan Raghuraman

Atul Luykx

Wei Dai

Hao Yang

Sunpreet Singh Arora

Mihai Christodorescu

Payman Mohassel

Mahdi Zamani

Minghua Xu

Yuhang Wu

Vinjith Nagaraja

Maliheh Shirvanian

Shengfei Gu

Karankumar Hiteshbhai Patel

Yizhen Wang

Ivan De Oliveira Nunes

Sikhar Patranabis

Shan Jin

Sebastian Meiser

Panagiotis Chatzigiannis

Harshal Bhadreshkumar Shah

Andrew Beams

Gayathri Annapurna Garimella

Methods and systems for securely generating secret shares in a distributed manner and distributing those secret shares to cryptographic devices are disclosed. The cryptographic devices can subsequently use these secret shares to perform threshold distributed cryptographic operations (such as encryption or decryption). A threshold number of generating cryptographic devices can each generate their own secret shares. These devices can also each generate partial secret shares that can be combined by receiving cryptographic devices to generate their own respective secret shares. Additionally, the generating devices can generate commitments corresponding to their secret shares. The generating devices can transmit the commitments to the other cryptographic devices and the partial secret shares to their corresponding receiving devices. At a later time, cryptographic devices possessing at least a threshold number of secret shares can collectively perform cryptographic operations using those secret shares.

Threshold secret share generation for distributed symmetric cryptography

Embodiments can perform a database join that allows secret shared database join between a database table with a unique matching column and a database table with a non-unique matching column (i.e., contains unbounded repeats of values).

Secret shared database joins using sorting

Methods and systems for performing federated private anomaly detection are disclosed. An anomaly detector computer can collaborate with an aggregator computer and an account management computer in order to train machine learning models, which can be used to classify events as not fraudulent or fraudulent. The anomaly detector computer can obliviously use private information (e.g., account flags and account flag values) held by the account management computer to train and use the machine learning models, such that the anomaly detector does not become aware of the account flags or account flags values. Such a system can be used, for example, for the detection and prevention of financial crime. The anomaly detector computer can use the account flag information possessed by the account management computer to identify fraudulent events performed by customers of the organization operating the account management computer.

Secure anomaly detection using federated learning

Secure protocols for external-facing authentication are provided for both user templates stored on their devices and the biometric measurement captured by external sensors of an access device. The protocols provide different levels of security, ranging from passive security with some leakage to active security with no leakage. A packing technique is also provided. Zero-knowledge techniques are used during enrollment to validate a norm of user templates and knowledge of the plaintext biometric template. Once enrolled, the verifier can sign the encrypted template for use in a later matching phase with an access device.

Verification of biometric templates for privacy preserving authentication

Provided is a method for authentication. The method may include receiving first password data, first biometric input data, and first secret key data. A second secret key may be generated based on the first secret key and the first password. Fuzzy extractor helper data may be generated based on the first biometric input and the second secret key. The fuzzy extractor helper data and the first secret key may be stored. The user may be authenticated based on an attempted password and/or a second biometric input. A system and computer program product are also disclosed.

Method, system, and computer program product for authentication

A homomorphic encryption scheme, such as Paillier encryption in combination with a bit packing process allows biometric matching at a terminal without exposing a biometric template stored at a user's device. Because such encryption schemes are data intensive, the bit packing process allows reductions in data being sent and processed so that the biometric matching process can be accomplished in near real time. The high speed of this optimized process allows the technique to be applied to many real world processes such as access control and transaction processing.

Optimized private biometric matching

A method performed by a user device is disclosed. The method comprising generating a secret and measuring a biometric template of a user operating the user device. The method then generates a plurality of secret shares of the secret and of the biometric template. The user device then transmits the secret shares of the secret and of the biometric template to a plurality of recovery devices. After, the user device may then initiate a recovery of the secret and measure a biometric measurement of the user. Data of the biometric measurement may be transmitted to the plurality of recovery devices, where the recovery devices perform a partial computation. The user device use the plurality of partial computations to determine a match between the biometric template and the biometric measurement. If the two biometrics match, the user device can reconstruct the secret using shares of the secret from the recovery devices.

Distributed private key recovery

Systems and methods for improved distributed symmetric cryptography are disclosed. A client computer may communicate with a number of cryptographic devices in order to encrypt or decrypt data. Each cryptographic device may possess a secret share and a verification share, which may be used in the process of encrypting or decrypting data. The client computer may generate a commitment and transmit the commitment to the cryptographic devices. Each cryptographic device may generate a partial computation based on the commitment and their respective secret share, and likewise generate a partial signature based on the commitment and their respective verification share. The partial computations and partial signatures may be transmitted to the client computer. The client computer may use the partial computations and partial signatures to generate a cryptographic key and verification signature respectively. The client computer may use the cryptographic key to encrypt or decrypt a message.

Distributed symmetric encryption

Systems, methods, and computing device readable media for implementing fast oblivious transfer between two computing devices may improve data security and computational efficiency. The various aspects may use random oracles with or without key agreements to improve the security of oblivious transfer key exchanges. Some techniques may include public/private key strategies for oblivious transfer, while other techniques may use key agreements to achieve simultaneous and efficient cryptographic key exchange.

Fast oblivious transfers

Secure protocols for external-facing authentication are provided for both user templates stored on their devices and the biometric measurement captured by external sensors of an access device. The protocols provide different levels of security, ranging from passive security with some leakage to active security with no leakage. A packing technique is also provided. Zero-knowledge techniques are used during enrollment to validate a norm of user templates and knowledge of the plaintext biometric template. One enrolled, the verifier can sign the encrypted template for use in a later matching phase with an access device.

Methods and systems according to embodiments of the invention provide for a framework for privacy-preserving machine learning which can be used to obtain solutions for training linear regression, logistic regression and neural network models. Embodiments of the invention are in a three-server model, wherein data owners secret-share their data among three servers who train and evaluate models on the joint data using three-party computation (3PC). Embodiments of the invention provide for efficient conversions between arithmetic, binary, and Yao 3PC, as well as techniques for fixed-point multiplication and truncation of shared decimal values. Embodiments also provide customized protocols for evaluating polynomial piecewise functions and a three-party oblivious transfer protocol.

Growing community of inventors

San Francisco, CA, United States of America

Peter Rindal