This work investigates the generalization behavior of deep neural networks (DNNs), focusing on the phenomenon of "fooling examples," where DNNs confidently classify inputs that appear random or unstructured to humans. To explore this phenomenon, we introduce an analytical framework based on maximum likelihood estimation (MLE), without adhering to conventional numerical approaches that rely on gradient-based optimization and explicit labels. Our analysis reveals that DNNs operating in an overparameterized regime exhibit a collapse in the output feature space. While this collapse improves network generalization, adding more layers eventually leads to a state of degeneracy, where the model learns trivial solutions by mapping distinct inputs to the same output, resulting in zero loss. Further investigation demonstrates that this degeneracy can be bypassed using our newly derived "wormhole" solution. The wormhole solution, when applied to arbitrary fooling examples, reconciles meaningful labels with random ones and provides a novel perspective on shortcut learning. These findings offer deeper insights into DNN generalization and highlight directions for future research on learning dynamics in unsupervised settings to bridge the gap between theory and practice.

The widespread use of cloud-based face recognition technology raises privacy concerns, as unauthorized access to face images can expose personal information or be exploited for fraudulent purposes. In response, privacy-preserving face recognition (PPFR) schemes have emerged to hide visual information and thwart unauthorized access. However, the validation methods employed by these schemes often rely on unrealistic assumptions, leaving doubts about their true effectiveness in safeguarding facial privacy. In this paper, we introduce a new approach to pri-vacy validation called Minimum Assumption Privacy Protection Validation (Map2 V). This is the first exploration of formulating a privacy validation method utilizing deep image priors and zeroth-order gradient estimation, with the potential to serve as a general framework for PPFR eval-uation. Building upon Map2v, we comprehensively vali-date the privacy-preserving capability of PPFRs through a combination of human and machine vision. The exper-iment results and analysis demonstrate the effectiveness and generalizability of the proposed Map2v, showcasing its superiority over native privacy validation methods from PPFR works of literature. Additionally, this work exposes privacy vulnerabilities in evaluated state-of-the-art P P FR schemes, laying the foundation for the subsequent effective proposal of countermeasures. The source code is available at https://github.com/Beauty9882/MAP2V.

Biometric cryptosystems present a promising avenue for secure authentication; however, the efficiency and security of such systems can be hindered by errors in biometric data. To address this challenge, existing systems employ error-correction codes, but often fail to consider the distribution of biometric sources, potentially leading to an underestimation of the system’s security. In response to this issue, we propose a novel algorithm pair, designated as ENCODE and DECODE, which facilitates direct codeword generation from biometric samples. Our approach accounts for the distribution of biometric sources, thereby providing a more accurate estimation of system security compared to traditional methods. Our proposed algorithm pair generates codewords that maintain interpretability and are sensitive to the cosine distance between original biometric samples. This similarity metric is particularly well-suited for high-dimensional data analysis and enables a precise assessment of system performance. We have rigorously established the correctness of our algorithm pair, and empirical results illustrate its efficacy in tolerating distance between codewords while preserving accuracy in cosine distance-sensitive contexts. This approach has the potential to significantly improve the efficiency and security of biometric cryptosystems, rendering them more appropriate for daily cryptographic applications.

Biometric Cryptosystems for secret binding such as fuzzy vault and fuzzy commitment are provable secure and offers a convenient way for secret management and protection. Despite numerous practical schemes have been reported, they are deficient in resisting several security and privacy attacks. In this paper, we propose a novel bio-cryptosystem that based on the three key ingredients namely Index of Maximum (IoM) hashing, (m, k) threshold secret sharing and b-band mini vaults notion. The IoM hashing is motivated from the ranking based Locality Sensitive Hashing theory meant for non-invertible transformation. On the other hand, the (m, k) threshold secret sharing scheme and the b-band mini vaults manage overcome inherent limitations of biometric cryptosystems when integrated with IoM hashing. The proposed scheme strikes the balance between performance and the privacy/security protection. Unlike fuzzy vault and fuzzy commitment, which primarily devised for unordered and binary biometrics, respectively, our scheme is tailored for feature vector-based biometrics (vectorial biometrics). Comprehensive experiments on fingerprint vectors that derived from several FVC fingerprint benchmarks and rigorous analysis demonstrate decent secret retrieval performance yet offer strong resilience against six major security and privacy attacks.

The rapid deployment of biometric authentication systems raises concern over user privacy and security. A biometric template protection scheme emerges as a solution to protect individual biometric templates stored in a database. Among all available protection schemes, a template protection scheme that relies on distance-preserving hashing has received much attention due to its simplicity and efficiency in offering privacy protection while archiving decent authentication performance. In this work, we introduce an efficient attack called known sample attack and demonstrate that most state-of-art template protection schemes that utilize distance-preserving hashing can be compromised in practice (within few seconds), especially when the output is significantly smaller than the original input sample size. These findings further motivated our subsequent work in proposing a secure authentication mechanism to resist such an attack with proper study over the distribution of the input samples. Furthermore, we conducted revocability, unlinkability analysis to demonstrate the satisfactory of general biometric template protection requirements; and showed the resistance of various security and privacy attacks, i.e., false acceptance attack, and attack via record multiplicity.

In this paper, we propose a ranking-based locality sensitive hashing inspired two-factor cancelable biometrics, dubbed “Index-of-Max” (IoM) hashing for biometric template protection. With externally generated random parameters, IoM hashing transforms a real-valued biometric feature vector into discrete index (max ranked) hashed code. We demonstrate two realizations from IoM hashing notion, namely, Gaussian random projection-based and uniformly random permutation-based hashing schemes. The discrete indices representation nature of IoM hashed codes enjoys several merits. First, IoM hashing empowers strong concealment to the biometric information. This contributes to the solid ground of non-invertibility guarantee. Second, IoM hashing is insensitive to the features magnitude, hence is more robust against biometric features variation. Third, the magnitude-independence trait of IoM hashing makes the hash codes being scale-invariant, which is critical for matching and feature alignment. The experimental results demonstrate favorable accuracy performance on benchmark FVC2002 and FVC2004 fingerprint databases. The analyses justify its resilience to the existing and newly introduced security and privacy attacks as well as satisfy the revocability and unlinkability criteria of cancelable biometrics.

This paper explores the recently published works on iris template protection namely Indexing-First-One hashing. Despite the Indexing-First-One hashing offers high recognition performance with resistant against several major privacy and security attacks, it does not resolve the rotation inconsistent issues existed in conventional iris template due to head tilt/rotation during user's eyes images acquisition. Hence, a pre-alignment step is required for the conventional IFO hashed code matching. Consequently, this increased the computational cost heavily. Hereby, we address the rotation inconsistent issue by proposing an alignment-free IFO hashing through a pre-transformation based on Bloom filter generation. The proposed pre-alignment IFO hashing shows promising recognition performance, and the pre-alignment procedure is eliminated to lower computational cost.