Federated learning is a decentralized machine learning technique that aggregates partial models trained by a set of clients on their own private data to obtain a global model. This technique is vulnerable to security attacks, such as model poisoning, whereby malicious clients submit bad updates in order to prevent the model from converging or to introduce artificial bias in the classification. Applying anti-poisoning techniques might lead to the discrimination of minority groups whose data are significantly and legitimately different from those of the majority of clients. In this work, we strive to strike a balance between fighting poisoning and accommodating diversity to help learning fairer and less discriminatory federated learning models. In this way, we forestall the exclusion of diverse clients while still ensuring detection of poisoning attacks. Empirical work on a standard machine learning data set shows that employing our approach to tell legitimate from malicious updates produces models that are more accurate than those obtained with standard poisoning detection techniques.

ABSTRACT Message encryption does not prevent eavesdroppers from unveiling who is communicating with whom, when, or how frequently, a privacy risk wireless networks are particularly vulnerable to. The Crowds protocol, a well‐established anonymous communication system, capitalizes on user collaboration to enforce sender anonymity. This work formulates a mathematical model of a Crowd‐like protocol for anonymous communication in a lossy network, establishes quantifiable metrics of anonymity and quality of service (QoS), and theoretically characterizes the trade‐off between them. The anonymity metric chosen follows the principle of measuring privacy as an attacker's estimation error. By introducing losses, we extend the applicability of the protocol beyond its original proposal. We quantify the intuition that anonymity comes at the expense of both delay and end‐to‐end losses. Aside from introducing losses in our model, another main difference with respect to the traditional Crowds is the focus on networks with stringent QoS requirements, for best effort anonymity, and the consequent elimination of the initial forwarding step. Beyond the mathematical solution, we illustrate a systematic methodology in our analysis of the protocol. This methodology includes a series of formal steps, from the establishment of quantifiable metrics all the way to the theoretical study of the privacy QoS trade‐off. Copyright © 2013 John Wiley & Sons, Ltd.

Collaborative tagging is one of the most popular services available online, and it allows end user to loosely classify either online or offline resources based on their feedback, expressed in the form of free-text labels (i.e., tags). Although tags may not be per se sensitive information, the wide use of collaborative tagging services increases the risk of cross referencing, thereby seriously compromising user privacy. In this paper, we make a first contribution toward the development of a privacy-preserving collaborative tagging service, by showing how a specific privacy-enhancing technology, namely tag suppression, can be used to protect end-user privacy. Moreover, we analyze how our approach can affect the effectiveness of a policy-based collaborative tagging system that supports enhanced web access functionalities, like content filtering and discovery, based on preferences specified by end users.

We address the problem of query profile obfuscation by means of partial query exchanges between two users, in order for their profiles of interest to appear distorted to the information provider (database, search engine, etc.). We illustrate a methodology to reach mutual privacy gain, that is, a situation where both users increase their own privacy protection through collaboration in query exchange. To this end, our approach starts with a mathematical formulation, involving the modeling of the users' apparent profiles as probability distributions over categories of interest, and the measure of their privacy as the corresponding Shannon entropy. The question of which query categories to exchange translates into finding optimization variables representing exchange policies, for various optimization objectives based on those entropies, possibly under exchange traffic constraints.

We present a mathematical formulation for the optimization of query forgery for private information retrieval, in the sense that the privacy risk is minimized for a given traffic and processing overhead. The privacy risk is measured as an information-theoretic divergence between the user's query distribution and the population's, which includes the entropy of the user's distribution as a special case. We carefully justify and interpret our privacy criterion from diverse perspectives. Our formulation poses a mathematically tractable problem that bears substantial resemblance with rate-distortion theory.

t-Closeness is a privacy model recently defined for data anonymization. A data set is said to satisfy t-closeness if, for each group of records sharing a combination of key attributes, the distance between the distribution of a confidential attribute in the group and the distribution of the attribute in the entire data set is no more than a threshold t. Here, we define a privacy measure in terms of information theory, similar to t-closeness. Then, we use the tools of that theory to show that our privacy measure can be achieved by the postrandomization method (PRAM) for masking in the discrete case, and by a form of noise addition in the general case.

We consider a first-order Markov source, which is predictively encoded using a DPCM-style encoder. The quantized compressed prediction residual is transmitted over an erasure channel. Additionally, a Wyner-Ziv encoded version of the prediction residual is transmitted in order to provide error resilience. When the symbols from the first transmission are erased by the channel, this second description is decoded, and limits the maximum distortion that can occur. Since the quantization step size used in the second description is, in general, larger than that used in the main transmission, error protection is lossy. Using high-rate quantization theory, we derive expressions for the rate and the end-to-end distortion incurred by this system. We show that, compared to conventional lossless forward error correction, this lossy error protection scheme is robust in the sense that it allows the received signal quality to degrade gracefully when the erasure probability increases

We extend the rate-distortion function for Wyner-Ziv coding of noisy sources with quadratic distortion, in the jointly Gaussian case, to more general statistics. It suffices that the noisy observation Z be the sum of a function of the side information Y and independent Gaussian noise, while the source data X must be the sum of a function of Y, a linear function of Z, and a random variable N such that the conditional expectation of N given Y and Z is zero, almost surely. Furthermore, the side information Y may be arbitrarily distributed in any alphabet, discrete or continuous. Under these general conditions, we prove that no rate loss is incurred due to the unavailability of the side information at the encoder. In the noiseless Wyner-Ziv case, i.e., when the source data is directly observed, the assumptions are still less restrictive than those recently established in the literature. We confirm, theoretically and experimentally, the consistency of this analysis with some of the main results on high-rate Wyner-Ziv quantization of noisy sources.

We investigate the design of optimal quantizers for individual encoding of several noisy observations of an unseen source, which is jointly decoded with the help of side information available at the decoder only. The joint statistics of the source data, the noisy observations and the side information are known, and exploited in the design. A variety of lossless coders for the quantization indices, including ideal Slepian-Wolf coders, are allowed. We present the optimality conditions such quantizers must satisfy, together with an extension of the Lloyd algorithm for a locally optimal design. Experimental results for Wyner-Ziv quantization of noisy Gaussian sources confirm the high-rate quantization theory established in our previous work.

Distributed coding is a new paradigm for video compression, based on Slepian and Wolf's and Wyner and Ziv's information-theoretic results from the 1970s. This paper reviews the recent development of practical distributed video coding schemes. Wyner-Ziv coding, i.e., lossy compression with receiver side information, enables low-complexity video encoding where the bulk of the computation is shifted to the decoder. Since the interframe dependence of the video sequence is exploited only at the decoder, an intraframe encoder can be combined with an interframe decoder. The rate-distortion performance is superior to conventional intraframe coding, but there is still a gap relative to conventional motion-compensated interframe coding. Wyner-Ziv coding is naturally robust against transmission errors and can be used for joint source-channel coding. A Wyner-Ziv MPEG encoder that protects the video waveform rather than the compressed bit stream achieves graceful degradation under deteriorating channel conditions without a layered signal representation.

The problem of designing optimal quantizers for distributed source coding is addressed. The generality of this formulation includes both the symmetric and asymmetric scenarios, together with a number of coding schemes, such as ideal coding achieving a rate equal to the joint conditional entropy of the quantized sources given the side information. The paper shows the optimality conditions that quantizers must satisfy, and generalizes the Lloyd algorithm for their design. Experimental results are shown for the Gaussian scalar asymmetric case.

A novel scheme for error-resilient digital video broadcasting, using Wyner-Ziv coding, is presented in this paper. We apply the general framework of systematic lossy source-channel coding to generate a supplementary bitstream that can correct transmission errors in the decoded video waveform up to a certain residual distortion. The systematic portion consists of a conventional MPEG-coded bitstream, which is transmitted over the error-prone channel without any forward error correction. The supplementary bitstream is a low rate representation of the transmitted video sequence generated using Wyner-Ziv encoding. We use the conventionally decoded error-concealed MPEG video sequence as side information to decode the Wyner-Ziv bits. The decoder combines the error-prone side information and the Wyner-Ziv description to yield an improved decoded video signal. We show how this scheme can be used to build a system which achieves graceful degradation without the need of a layered representation of the video waveform.