In certain specific scenarios, there is a risk of privacy leakage in terms of the soft biometric attributes on a person’s face. However, existing face privacy-enhancing techniques suffer from limited generalizability, meaning that they can only induce misclassification in a specific classifier but fail to generalize this effect well to arbitrary attribute classifiers. Moreover, existing methods reverse attributes to improve face privacy, but this may result in privacy recovery. To address those problems, we propose GFPNet, a novel privacy-enhancing model that can provide generalizable and reliable privacy to face images. The key factor for improving generalizability is that GFPNet uses defense training, which is an effective way to improve model robustness, to dynamically strengthen the mediocre auxiliary attribute classifier during iterative training. Specifically, the generalizability of GFPNet is enhanced in the game between attack and defense, where the generator attempts to deceive the auxiliary attribute classifier and the classifier defends against the generator’s attack by defense training. Furthermore, instead of reversing attributes, skewing attributes to one side is used to avoid attribute recovery. GFPNet also integrates a face matcher, multi-scale discriminator, and Demiguise Attack to improve face matching and image quality. Extensive experiments demonstrate that GFPNet has excellent generalizability to arbitrary attribute classifiers and satisfactory face-matching utility.

Distributed data storage (DDS) in mobile crowdsensing (MCS) systems has recently gained popularity. Data should be briefly saved on participants’ mobile devices before being gathered once the centralized cloud servers resume normal operations. For MCS systems, the existing DDS strategies briefly considered reconstructing the scene as precisely as possible without thinking about the costs of each step. However, our goal is to obtain a sufficiently accurate approximation of the sensing data from mobile participants with as few costs as possible. We note a crucial observation: when a specified number of participants have been transmitted to a central server, the sensing data has already been well reconstructed, and the accuracy advancement with additional transmitted participants is minimal. In our scheme, two stopping criteria are proposed for DDS in compressive MCS, which aims to enhance recovery performance while reducing the costs of the whole process. In the first stopping criterion, we established a rule to stop the continued recruitment of participants. The algorithm adaptively increases the number of participants until the reconstruction accuracy meets the requirement. Another stopping criterion of the reconstruction algorithm is designed to find a more accurate number of iterations than the original. The experiment results demonstrate that the first stopping criterion can reduce participants’ collection while obtaining an approximate value. The second stopping criterion assists the reconstruction algorithm in terminating at a more appropriate number of iterations, saving computing costs while ensuring accuracy.

Privacy concerns over face recognition systems have attracted extensive attention in various fields. For gender privacy-preserving work, there are two key challenges: 1)privacy, i.e., confusing gender classifiers and 2)utility, i.e., maintaining its face verification performance. To address both issues, this article develops a novel gender-adversarial network, referred to as Gender-AN, to impart gender privacy to face images. Gender-AN employs an attribute-independent encoder–decoder GAN-based network to perturb the input face image, training with the assistance of the proper facial attributes. The perturbed image is then able to obfuscate gender classifiers while maintaining identity discriminability. To optimize the generator, a multitask-based loss function is utilized, which includes attribute manipulation loss, face matcher loss, adversarial loss, and reconstruction loss functions. This optimization facilitates our model to achieve the generalization, verification preserve, and natural appearance, simultaneously. Extensive experiments confirm the effectiveness of the proposed model in enhancing gender privacy and preserving face verification utility.

Collaborative representation classification (CRC) has attracted increasing attention in face recognition (FR) tasks. The two-phase sparse representation (TPSR) methods are the improved schemes. However, most TPSR methods decrease training samples in the first step, resulting in less similarities or discrimination for representation, even unstable classification. In this paper, we propose a new two-phase representation based FR approach with random-filtering virtual samples, called Random-Filtering based Sparse Representation (RFSR) scheme. To increase the similarity in the same class and the discrimination between different classes, RFSR first uses original training samples and their corresponding random-filtering virtual samples to constructs a new training set. Then it exploits the new training set to perform CRC. The experiment results indicate that our method outperforms the two-phase test sample sparse representation (TPTSSR) method and the simple and fast representation-based (SFRB) scheme.