Keyless claims first as split biometric hashing wins international patent

International companies including Keyless, Smart Engines and Trust Stamp, and academic researchers are securing patents for a host of AI-based applications, including biometric verification systems.

Keyless system splits biometric hashing between device and the cloud

In a Biometric Update exclusive, London-based biometrics and passwordless authentication firm Keyless announces that its Biometric Authentication Key Exchange (BAKE) technology has been approved for a patent under the World Intellectual Property Organization’s Patent Cooperation Treaty. With BAKE, the biometric data hashing process is split between the user’s device and a cloud server, ensuring that biometric templates never fully leave the user’s device in their raw form.

The announcement from Keyless says the system addresses evolving data privacy regulation and compliance, as well as the long-standing security issues in the use of biometric hashes in authentication.

Hashes can be reverse-engineered to reveal the original templates from which they were generated. Splitting the hashing process between the device and cloud server – the first system to do so, the company says – minimizes the risks of data compromise, since the fully hashed data undergoes a cryptographic protocol with the cloud server, which returns a secret key but retains nothing from the transaction.

“We are constantly pushing the boundaries of biometric authentication technology to create more secure and privacy-respecting solutions,” says Keyless CTO Paolo Gasti. “BAKE is a testament to our commitment to innovation, compliance, and user privacy. With this new innovation, we now offer a superior biometric authentication method that combines the benefits of local biometrics with the security of a cryptographic protocol.”

Keyless says BAKE complements its existing suite of hashed selfie-based Zero-Knowledge Biometric technologies, and advances the firm’s core commitment to user privacy.

Smart Engines mixes neural networks and classic algorithm

Armenian firm Smart Engines, which makes software for computer vision, ID document processing and optical character recognition, issued a release announcing a new U.S. patent for an image analysis and modification tool that uses AI to “restore the parameters of three-dimensional space from a two-dimensional image.”

Smart Engines’ system builds a layered neural network atop the Hough Transform, a classic algorithm that detects lines and geometric shapes in photographic images and uses that data to predict the linear elements and contours of incomplete, obscured or distorted shapes. Hough Transforms can search and highlight straight lines of text or the boundaries of objects such as ID documents. By combining Hough Transforms with the convolutional layers of a neural network, the system can detect local features from image points, run the data through the HT to map those values along straight lines, then leverage the neural network to transpose the local feature maps to the original coordinates.

“At the last stage,” explains Alexander Sheshkus, Smart Engines’ head of machine learning, “the obtained result is processed by another block of ultra-precise layers, as a result of which we receive an image in the original coordinates, but at each point information from the entire image has been accumulated.”

For the ID verification space, a key takeaway is that AI can restore partially obscured images – for instance, the edge of an ID document blocked by a finger. The system’s robust object detection capability thus makes it resistant to tampering and fraud attacks.

The HT-neural network combination is Smart Engines’ sixth U.S. patent.

Trust Stamp gets notice of allowance for new encryption patent application

In keeping with its recent trend, Atlanta-based biometric digital identity services firm Trust Stamp has secured a notice of allowance from the U.S. Patent and Trademark Office for utility patent application 17/401,504, covering “Systems and Methods for Liveness-Verified, Biometric Based Encryption.”

In a press release, Trust Stamp’s Chief Science Officer Dr. Norman Poh noted that since the rapid acceleration of artificial intelligence tools will increase the capabilities of attackers, AI-powered defenses must stay one step ahead.

“Our liveness detection technology used in conjunction with our patented identity tokenization is designed to defend against the escalating occurrence of presentation attacks,” he says, noting JP Morgan’s recent statement in Davos that it is fighting off around 45 billion infiltration attempts per day.

A pair of scientists notch biometric wins

A Saudi Arabia-based team from King Saud University, led by Pakistani scientist Dr. Muhammad Khurram Khan, has been granted a patent for a “biometrics security invention,” according to Pakistan Today. The work reportedly focuses on advancing biometric identification methods using AI.

The biometrics patent is the second awarded to Khan in under a year.

And a press release from the University of South Florida says Professor Sudeep Sarkar has been named a Distinguished Member of the Association for Computing Machinery (ACM), for contributions to the advancement of computer vision and gait biometrics.

Article Topics

biometric hash  |  biometrics  |  Keyless  |  patents  |  research and development  |  Smart Engines  |  Trust Stamp