Privacy-Preserving Image Template Sharing Using Contrastive Learning.

With the recent developments of Machine Learning as a Service (MLaaS), various privacy concerns have been raised. Having access to the user's data, an adversary can design attacks with different objectives, namely, reconstruction or attribute inference attacks. In this paper, we propose two different training frameworks for an image classification task while preserving user data privacy against the two aforementioned attacks. In both frameworks, an encoder is trained with contrastive loss, providing a superior utility-privacy trade-off. In the reconstruction attack scenario, a supervised contrastive loss was employed to provide maximal discrimination for the targeted classification task. The encoded features are further perturbed using the obfuscator module to remove all redundant information. Moreover, the obfuscator module is jointly trained with a classifier to minimize the correlation between private feature representation and original data while retaining the model utility for the classification. For the attribute inference attack, we aim to provide a representation of data that is independent of the sensitive attribute. Therefore, the encoder is trained with supervised and private contrastive loss. Furthermore, an obfuscator module is trained in an adversarial manner to preserve the privacy of sensitive attributes while maintaining the classification performance on the target attribute. The reported results on the CelebA dataset validate the effectiveness of the proposed frameworks.

Information Bottleneck Classification in Extremely Distributed Systems.

We present a new decentralized classification system based on a distributed architecture. This system consists of distributed nodes, each possessing their own datasets and computing modules, along with a centralized server, which provides probes to classification and aggregates the responses of nodes for a final decision. Each node, with access to its own training dataset of a given class, is trained based on an auto-encoder system consisting of a fixed data-independent encoder, a pre-trained quantizer and a class-dependent decoder. Hence, these auto-encoders are highly dependent on the class probability distribution for which the reconstruction distortion is minimized. Alternatively, when an encoding-quantizing-decoding node observes data from different distributions, unseen at training, there is a mismatch, and such a decoding is not optimal, leading to a significant increase of the reconstruction distortion. The final classification is performed at the centralized classifier that votes for the class with the minimum reconstruction distortion. In addition to the system applicability for applications facing big-data communication problems and or requiring private classification, the above distributed scheme creates a theoretical bridge to the information bottleneck principle. The proposed system demonstrates a very promising performance on basic datasets such as MNIST and FasionMNIST.

Machine learning through cryptographic glasses: combating adversarial attacks by key-based diversified aggregation.

In recent years, classification techniques based on deep neural networks (DNN) were widely used in many fields such as computer vision, natural language processing, and self-driving cars. However, the vulnerability of the DNN-based classification systems to adversarial attacks questions their usage in many critical applications. Therefore, the development of robust DNN-based classifiers is a critical point for the future deployment of these methods. Not less important issue is understanding of the mechanisms behind this vulnerability. Additionally, it is not completely clear how to link machine learning with cryptography to create an information advantage of the defender over the attacker. In this paper, we propose a key-based diversified aggregation (KDA) mechanism as a defense strategy in a gray- and black-box scenario. KDA assumes that the attacker (i) knows the architecture of classifier and the used defense strategy, (ii) has an access to the training data set, but (iii) does not know a secret key and does not have access to the internal states of the system. The robustness of the system is achieved by a specially designed key-based randomization. The proposed randomization prevents the gradients' back propagation and restricts the attacker to create a "bypass" system. The randomization is performed simultaneously in several channels. Each channel introduces its own randomization in a special transform domain. The sharing of a secret key between the training and test stages creates an information advantage to the defender. Finally, the aggregation of soft outputs from each channel stabilizes the results and increases the reliability of the final score. The performed experimental evaluation demonstrates a high robustness and universality of the KDA against state-of-the-art gradient-based gray-box transferability attacks and the non-gradient-based black-box attacks (The results reported in this paper have been partially presented in CVPR 2019 (Taran et al., Defending against adversarial attacks by randomized diversification, 2019) & ICIP 2019 (Taran et al., Robustification of deep net classifiers by key-based diversified aggregation with pre-filtering, 2019)).