This paper introduces a novel protocol for privacy-preserving biometric identification, named Monchi, that combines the use of homomorphic encryption for the computation of the identification score with function secret sharing to obliviously compare this score with a given threshold and finally output the binary result. Given the cost of homomorphic encryption, BFV in this solution, we study and evaluate the integration of two packing solutions that enable the regrouping of multiple templates in one ciphertext to improve efficiency meaningfully. We propose an end-to-end protocol, prove it secure and implement it. Our experimental results attest to Monchi's applicability to the real-life use case of an airplane boarding scenario with 1000 passengers,taking less than one second to authorize/deny access to the plane to each passenger via biometric identification while maintaining the privacy of all passengers.

This paper proposes a novel method to perform privacy-preserving face identification based on the notion of group testing, and applies it to a solution using the Cheon-Kim-Kim-Song (CKKS) homomorphic encryption scheme. Securely computing the closest reference template to a given live template requires K comparisons, as many as there are identities in a biometric database. Our solution, named Grote, replaces element-wise testing by group testing to drastically reduce the number of such costly, non-linear operations in the encrypted domain from K to up to 2\sqrtK . More specifically, we approximate the max of the coordinates of a large vector by raising to the α-th power and cumulative sum in a 2D layout, incurring a small impact in the accuracy of the system while greatly speeding up its execution. We implement Grote and evaluate its performance.

We introduce a new wallet recovery pro-cess. Our solution associates 1) visual passwords: a photograph ofa secretly picked object (Chabanne et aI., 2013) with 2) ImageNet classifiers transforming images into binary vectors and, 3) obfuscated fuzzy matching (Galbraith and Zobernig, 2019) for the storage of visual passwords/retrieval of wallet seeds. Our experiments show that the replacement of long seed phrases by a photograph is possible.

In spite of the high performance and reliability of deep learning algorithms in a wide range of everyday applications, many investigations tend to show that a lot of models exhibit biases, discriminating against specific subgroups of the population (e.g. gender, ethnicity). This urges the practitioner to develop fair systems with a uniform/comparable performance across sensitive groups. In this work, we investigate the gender bias of deep Face Recognition networks. In order to measure this bias, we introduce two new metrics, BFAR and BFRR, that better reflect the inherent deployment needs of Face Recognition systems. Motivated by geometric considerations, we mitigate gender bias through a new post-processing methodology which transforms the deep embeddings of a pre-trained model to give more representation power to discriminated subgroups. It consists in training a shallow neural network by minimizing a Fair von Mises-Fisher loss whose hyperparameters account for the intra-class variance of each gender. Interestingly, we empirically observe that these hyperparameters are correlated with our fairness metrics. In fact, extensive numerical experiments on a variety of datasets show that a careful selection significantly reduces gender bias.

This paper proposes a novel collaborative decryption protocol for the Brakerski-Fan-Vercauteren (BFV) homomorphic encryption scheme in a multiparty distributed setting, and puts it to use in designing a leakage-resilient biometric identification solution. Allowing the computation of standard homomorphic operations over encrypted data, our protocol reveals only one least significant bit (LSB) of a scalar/vectorized result resorting to a pool of N parties. By employing additively shared masking, our solution preserves the privacy of all the remaining bits in the result as long as one party remains honest. We formalize the protocol, prove it secure in several adversarial models, implement it on top of the open-source library Lattigo and showcase its applicability as part of a biometric access control scenario.

Given oracle access to a Neural Network (NN), it is possible to extract its underlying model. We here introduce a protection by adding parasitic layers which keep the underlying NN's predictions mostly unchanged while complexifying the task of reverse-engineering. Our countermeasure relies on approximating a noisy identity mapping with a Convolutional NN. We explain why the introduction of new parasitic layers complexifies the attacks. We report experiments regarding the performance and the accuracy of the protected NN.