arXiv subjects

Pablo Piantanida

Publications and source records attributed to Pablo Piantanida.

At least 73 records · Page 4Linked to original sources

arXiv2022

Igeood: An Information Geometry Approach to Out-of-Distribution Detection

Reliable out-of-distribution (OOD) detection is fundamental to implementing safer modern machine learning (ML) systems. In this paper, we introduce Igeood, an effective method for detecting OOD samples. Igeood applies to any pre-trained neural network, works under various degrees of access to the ML model, does not require OOD samples or assumptions on the OOD data but can also benefit (if available) from OOD samples. By building on the geodesic (Fisher-Rao) distance between the underlying data distributions, our discriminator can combine confidence scores from the logits outputs and the learned features of a deep neural network. Empirically, we show that Igeood outperforms competing state-of-the-art methods on a variety of network architectures and datasets.

stat.ML↗

arXiv2021

PACMAN: PAC-style bounds accounting for the Mismatch between Accuracy and Negative log-loss

The ultimate performance of machine learning algorithms for classification tasks is usually measured in terms of the empirical error probability (or accuracy) based on a testing dataset. Whereas, these algorithms are optimized through the minimization of a typically different--more convenient--loss function based on a training set. For classification tasks, this loss function is often the negative log-loss that leads to the well-known cross-entropy risk which is typically better behaved (from a numerical perspective) than the error probability. Conventional studies on the generalization error do not usually take into account the underlying mismatch between losses at training and testing phases. In this work, we introduce an analysis based on point-wise PAC approach over the generalization gap considering the mismatch of testing based on the accuracy metric and training on the negative log-loss. We label this analysis PACMAN. Building on the fact that the mentioned mismatch can be written as a likelihood ratio, concentration inequalities can be used to provide some insights for the generalization problem in terms of some point-wise PAC bounds depending on some meaningful information-theoretic quantities. An analysis of the obtained bounds and a comparison with available results in the literature are also provided.

cs.LG↗

arXiv2021

A unifying mutual information view of metric learning: cross-entropy vs. pairwise losses

Recently, substantial research efforts in Deep Metric Learning (DML) focused on designing complex pairwise-distance losses, which require convoluted schemes to ease optimization, such as sample mining or pair weighting. The standard cross-entropy loss for classification has been largely overlooked in DML. On the surface, the cross-entropy may seem unrelated and irrelevant to metric learning as it does not explicitly involve pairwise distances. However, we provide a theoretical analysis that links the cross-entropy to several well-known and recent pairwise losses. Our connections are drawn from two different perspectives: one based on an explicit optimization insight; the other on discriminative and generative views of the mutual information between the labels and the learned features. First, we explicitly demonstrate that the cross-entropy is an upper bound on a new pairwise loss, which has a structure similar to various pairwise losses: it minimizes intra-class distances while maximizing inter-class distances. As a result, minimizing the cross-entropy can be seen as an approximate bound-optimization (or Majorize-Minimize) algorithm for minimizing this pairwise loss. Second, we show that, more generally, minimizing the cross-entropy is actually equivalent to maximizing the mutual information, to which we connect several well-known pairwise losses. Furthermore, we show that various standard pairwise losses can be explicitly related to one another via bound relationships. Our findings indicate that the cross-entropy represents a proxy for maximizing the mutual information -- as pairwise losses do -- without the need for convoluted sample-mining heuristics. Our experiments over four standard DML benchmarks strongly support our findings. We obtain state-of-the-art results, outperforming recent and complex DML methods.

cs.LG↗

arXiv2021

Universal Weak Variable-Length Source Coding on Countable Infinite Alphabets

Motivated from the fact that universal source coding on countably infinite alphabets is not feasible, this work introduces the notion of almost lossless source coding. Analog to the weak variable-length source coding problem studied by Han (IEEE TIT, 2000, 46, 1217-1226), almost lossless source coding aims at relaxing the lossless block-wise assumption to allow an average per-letter distortion that vanishes asymptotically as the block-length tends to infinity. In this setup, we show on one hand that Shannon entropy characterizes the minimum achievable rate (similarly to the case of finite alphabet sources) while on the other that almost lossless universal source coding becomes feasible for the family of finite-entropy stationary memoryless sources with infinite alphabets. Furthermore, we study a stronger notion of almost lossless universality that demands uniform convergence of the average per-letter distortion to zero, where we establish a necessary and sufficient condition for the so-called family of envelope distributions to achieve it. Remarkably, this condition is the same necessary and sufficient condition needed for the existence of a strongly minimax (lossless) universal source code for the family of envelope distributions. Finally, we show that an almost lossless coding scheme offers faster rate of convergence for the (minimax) redundancy compared to the well-known information radius developed for the lossless case at the expense of tolerating a non-zero distortion that vanishes to zero as the block-length grows. This shows that even when lossless universality is feasible, an almost lossless scheme can offer different regimes on the rates of convergence of the (worst case) redundancy versus the (worst case) distortion.

cs.IT↗

arXiv2021

Privacy-Preserving Synthetic Smart Meters Data

Power consumption data is very useful as it allows to optimize power grids, detect anomalies and prevent failures, on top of being useful for diverse research purposes. However, the use of power consumption data raises significant privacy concerns, as this data usually belongs to clients of a power company. As a solution, we propose a method to generate synthetic power consumption samples that faithfully imitate the originals, but are detached from the clients and their identities. Our method is based on Generative Adversarial Networks (GANs). Our contribution is twofold. First, we focus on the quality of the generated data, which is not a trivial task as no standard evaluation methods are available. Then, we study the privacy guarantees provided to members of the training set of our neural network. As a minimum requirement for privacy, we demand our neural network to be robust to membership inference attacks, as these provide a gateway for further attacks in addition to presenting a privacy threat on their own. We find that there is a compromise to be made between the privacy and the performance provided by the algorithm.

eess.SP↗

arXiv2021

Distributed Information-Theoretic Clustering

We study a novel multi-terminal source coding setup motivated by the biclustering problem. Two separate encoders observe two i.i.d. sequences $X^n$ and $Y^n$, respectively. The goal is to find rate-limited encodings $f(x^n)$ and $g(z^n)$ that maximize the mutual information $I(f(X^n); g(Y^n))/n$. We discuss connections of this problem with hypothesis testing against independence, pattern recognition, and the information bottleneck method. Improving previous cardinality bounds for the inner and outer bounds allows us to thoroughly study the special case of a binary symmetric source and to quantify the gap between the inner and the outer bound in this special case. Furthermore, we investigate a multiple description (MD) extension of the Chief Operating Officer (CEO) problem with mutual information constraint. Surprisingly, this MD-CEO problem permits a tight single-letter characterization of the achievable region.

cs.IT↗

arXiv2021

Real-Time Privacy-Preserving Data Release for Smart Meters

Smart Meters (SMs) are able to share the power consumption of users with utility providers almost in real-time. These fine-grained signals carry sensitive information about users, which has raised serious concerns from the privacy viewpoint. In this paper, we focus on real-time privacy threats, i.e., potential attackers that try to infer sensitive information from SMs data in an online fashion. We adopt an information-theoretic privacy measure and show that it effectively limits the performance of any attacker. Then, we propose a general formulation to design a privatization mechanism that can provide a target level of privacy by adding a minimal amount of distortion to the SMs measurements. On the other hand, to cope with different applications, a flexible distortion measure is considered. This formulation leads to a general loss function, which is optimized using a deep learning adversarial framework, where two neural networks -- referred to as the releaser and the adversary -- are trained with opposite goals. An exhaustive empirical study is then performed to validate the performance of the proposed approach and compare it with state-of-the-art methods for the occupancy detection privacy problem. Finally, we also investigate the impact of data mismatch between the releaser and the attacker.

eess.SP↗

arXiv2021

On Direct Distribution Matching for Adapting Segmentation Networks

Minimization of distribution matching losses is a principled approach to domain adaptation in the context of image classification. However, it is largely overlooked in adapting segmentation networks, which is currently dominated by adversarial models. We propose a class of loss functions, which encourage direct kernel density matching in the network-output space, up to some geometric transformations computed from unlabeled inputs. Rather than using an intermediate domain discriminator, our direct approach unifies distribution matching and segmentation in a single loss. Therefore, it simplifies segmentation adaptation by avoiding extra adversarial steps, while improving both the quality, stability and efficiency of training. We juxtapose our approach to state-of-the-art segmentation adaptation via adversarial training in the network-output space. In the challenging task of adapting brain segmentation across different magnetic resonance images (MRI) modalities, our approach achieves significantly better results both in terms of accuracy and stability.

cs.CV↗

arXiv2021

On the Exponential Approximation of Type II Error Probability of Distributed Test of Independence

This paper studies distributed binary test of statistical independence under communication (information bits) constraints. While testing independence is very relevant in various applications, distributed independence test is particularly useful for event detection in sensor networks where data correlation often occurs among observations of devices in the presence of a signal of interest. By focusing on the case of two devices because of their tractability, we begin by investigating conditions on Type I error probability restrictions under which the minimum Type II error admits an exponential behavior with the sample size. Then, we study the finite sample-size regime of this problem. We derive new upper and lower bounds for the gap between the minimum Type II error and its exponential approximation under different setups, including restrictions imposed on the vanishing Type I error probability. Our theoretical results shed light on the sample-size regimes at which approximations of the Type II error probability via error exponents became informative enough in the sense of predicting well the actual error probability. We finally discuss an application of our results where the gap is evaluated numerically, and we show that exponential approximations are not only tractable but also a valuable proxy for the Type II probability of error in the finite-length regime.

math.ST↗

arXiv2021

On the Estimation of Information Measures of Continuous Distributions

The estimation of information measures of continuous distributions based on samples is a fundamental problem in statistics and machine learning. In this paper, we analyze estimates of differential entropy in $K$-dimensional Euclidean space, computed from a finite number of samples, when the probability density function belongs to a predetermined convex family $\mathcal{P}$. First, estimating differential entropy to any accuracy is shown to be infeasible if the differential entropy of densities in $\mathcal{P}$ is unbounded, clearly showing the necessity of additional assumptions. Subsequently, we investigate sufficient conditions that enable confidence bounds for the estimation of differential entropy. In particular, we provide confidence bounds for simple histogram based estimation of differential entropy from a fixed number of samples, assuming that the probability density function is Lipschitz continuous with known Lipschitz constant and known, bounded support. Our focus is on differential entropy, but we provide examples that show that similar results hold for mutual information and relative entropy as well.

cs.IT↗

arXiv2021

Estimating g-Leakage via Machine Learning

This paper considers the problem of estimating the information leakage of a system in the black-box scenario. It is assumed that the system's internals are unknown to the learner, or anyway too complicated to analyze, and the only available information are pairs of input-output data samples, possibly obtained by submitting queries to the system or provided by a third party. Previous research has mainly focused on counting the frequencies to estimate the input-output conditional probabilities (referred to as frequentist approach), however this method is not accurate when the domain of possible outputs is large. To overcome this difficulty, the estimation of the Bayes error of the ideal classifier was recently investigated using Machine Learning (ML) models and it has been shown to be more accurate thanks to the ability of those models to learn the input-output correspondence. However, the Bayes vulnerability is only suitable to describe one-try attacks. A more general and flexible measure of leakage is the g-vulnerability, which encompasses several different types of adversaries, with different goals and capabilities. In this paper, we propose a novel approach to perform black-box estimation of the g-vulnerability using ML. A feature of our approach is that it does not require to estimate the conditional probabilities, and that it is suitable for a large class of ML algorithms. First, we formally show the learnability for all data distributions. Then, we evaluate the performance via various experiments using k-Nearest Neighbors and Neural Networks. Our results outperform the frequentist approach when the observables domain is large.

cs.CR↗

arXiv2021

Privacy-Cost Management in Smart Meters Using Deep Reinforcement Learning

Smart meters (SMs) play a pivotal rule in the smart grid by being able to report the electricity usage of consumers to the utility provider (UP) almost in real-time. However, this could leak sensitive information about the consumers to the UP or a third-party. Recent works have leveraged the availability of energy storage devices, e.g., a rechargeable battery (RB), in order to provide privacy to the consumers with minimal additional energy cost. In this paper, a privacy-cost management unit (PCMU) is proposed based on a model-free deep reinforcement learning algorithm, called deep double Q-learning (DDQL). Empirical results evaluated on actual SMs data are presented to compare DDQL with the state-of-the-art, i.e., classical Q-learning (CQL). Additionally, the performance of the method is investigated for two concrete cases where attackers aim to infer the actual demand load and the occupancy status of dwellings. Finally, an abstract information-theoretic characterization is provided.

eess.SP↗

arXiv2021

Finite-Length Bounds on Hypothesis Testing Subject to Vanishing Type I Error Restrictions

A central problem in Binary Hypothesis Testing (BHT) is to determine the optimal tradeoff between the Type I error (referred to as false alarm) and Type II (referred to as miss) error. In this context, the exponential rate of convergence of the optimal miss error probability -- as the sample size tends to infinity -- given some (positive) restrictions on the false alarm probabilities is a fundamental question to address in theory. Considering the more realistic context of a BHT with a finite number of observations, this paper presents a new non-asymptotic result for the scenario with monotonic (sub-exponential decreasing) restriction on the Type I error probability, which extends the result presented by Strassen in 2009. Building on the use of concentration inequalities, we offer new upper and lower bounds to the optimal Type II error probability for the case of finite observations. Finally, the derived bounds are evaluated and interpreted numerically (as a function of the number samples) for some vanishing Type I error restrictions.

cs.IT↗

arXiv2021

Deep Directed Information-Based Learning for Privacy-Preserving Smart Meter Data Release

The explosion of data collection has raised serious privacy concerns in users due to the possibility that sharing data may also reveal sensitive information. The main goal of a privacy-preserving mechanism is to prevent a malicious third party from inferring sensitive information while keeping the shared data useful. In this paper, we study this problem in the context of time series data and smart meters (SMs) power consumption measurements in particular. Although Mutual Information (MI) between private and released variables has been used as a common information-theoretic privacy measure, it fails to capture the causal time dependencies present in the power consumption time series data. To overcome this limitation, we introduce the Directed Information (DI) as a more meaningful measure of privacy in the considered setting and propose a novel loss function. The optimization is then performed using an adversarial framework where two Recurrent Neural Networks (RNNs), referred to as the releaser and the adversary, are trained with opposite goals. Our empirical studies on real-world data sets from SMs measurements in the worst-case scenario where an attacker has access to all the training data set used by the releaser, validate the proposed method and show the existing trade-offs between privacy and utility.

cs.LG↗

arXiv2021

On the Impact of Side Information on Smart Meter Privacy-Preserving Methods

Smart meters (SMs) can pose privacy threats for consumers, an issue that has received significant attention in recent years. This paper studies the impact of Side Information (SI) on the performance of distortion-based real-time privacy-preserving algorithms for SMs. In particular, we consider a deep adversarial learning framework, in which the desired releaser (a recurrent neural network) is trained by fighting against an adversary network until convergence. To define the loss functions, two different approaches are considered: the Causal Adversarial Learning (CAL) and the Directed Information (DI)-based learning. The main difference between these approaches is in how the privacy term is measured during the training process. On the one hand, the releaser in the CAL method, by getting supervision from the actual values of the private variables and feedback from the adversary performance, tries to minimize the adversary log-likelihood. On the other hand, the releaser in the DI approach completely relies on the feedback received from the adversary and is optimized to maximize its uncertainty. The performance of these two algorithms is evaluated empirically using real-world SMs data, considering an attacker with access to SI (e.g., the day of the week) that tries to infer the occupancy status from the released SMs data. The results show that, although they perform similarly when the attacker does not exploit the SI, in general, the CAL method is less sensitive to the inclusion of SI. However, in both cases, privacy levels are significantly affected, particularly when multiple sources of SI are included.

eess.SP↗

arXiv2021

Collaborative Information Bottleneck

This paper investigates a multi-terminal source coding problem under a logarithmic loss fidelity which does not necessarily lead to an additive distortion measure. The problem is motivated by an extension of the Information Bottleneck method to a multi-source scenario where several encoders have to build cooperatively rate-limited descriptions of their sources in order to maximize information with respect to other unobserved (hidden) sources. More precisely, we study fundamental information-theoretic limits of the so-called: (i) Two-way Collaborative Information Bottleneck (TW-CIB) and (ii) the Collaborative Distributed Information Bottleneck (CDIB) problems. The TW-CIB problem consists of two distant encoders that separately observe marginal (dependent) components $X_1$ and $X_2$ and can cooperate through multiple exchanges of limited information with the aim of extracting information about hidden variables $(Y_1,Y_2)$, which can be arbitrarily dependent on $(X_1,X_2)$. On the other hand, in CDIB there are two cooperating encoders which separately observe $X_1$ and $X_2$ and a third node which can listen to the exchanges between the two encoders in order to obtain information about a hidden variable $Y$. The relevance (figure-of-merit) is measured in terms of a normalized (per-sample) multi-letter mutual information metric (log-loss fidelity) and an interesting tradeoff arises by constraining the complexity of descriptions, measured in terms of the rates needed for the exchanges between the encoders and decoders involved. Inner and outer bounds to the complexity-relevance region of these problems are derived from which optimality is characterized for several cases of interest. Our resulting theoretical complexity-relevance regions are finally evaluated for binary symmetric and Gaussian statistical models.

cs.IT↗

arXiv2021

On the impossibility of non-trivial accuracy under fairness constraints

One of the main concerns about fairness in machine learning (ML) is that, in order to achieve it, one may have to trade off some accuracy. To overcome this issue, Hardt et al. proposed the notion of equality of opportunity (EO), which is compatible with maximal accuracy when the target label is deterministic with respect to the input features. In the probabilistic case, however, the issue is more complicated: It has been shown that under differential privacy constraints, there are data sources for which EO can only be achieved at the total detriment of accuracy, in the sense that a classifier that satisfies EO cannot be more accurate than a trivial (i.e., constant) classifier. In our paper we strengthen this result by removing the privacy constraint. Namely, we show that for certain data sources, the most accurate classifier that satisfies EO is a trivial classifier. Furthermore, we study the trade-off between accuracy and EO loss (opportunity difference), and provide a sufficient condition on the data source under which EO and non-trivial accuracy are compatible.

cs.LG↗

arXiv2021

Privacy-Cost Management in Smart Meters with Mutual Information-Based Reinforcement Learning

The rapid development and expansion of the Internet of Things (IoT) paradigm has drastically increased the collection and exchange of data between sensors and systems, a phenomenon that raises serious privacy concerns in some domains. In particular, Smart Meters (SMs) share fine-grained electricity consumption of households with utility providers that can potentially violate users' privacy as sensitive information is leaked through the data. In order to enhance privacy, the electricity consumers can exploit the availability of physical resources such as a rechargeable battery (RB) to shape their power demand as dictated by a Privacy-Cost Management Unit (PCMU). In this paper, we present a novel method to learn the PCMU policy using Deep Reinforcement Learning (DRL). We adopt the mutual information (MI) between the user's demand load and the masked load seen by the power grid as a reliable and general privacy measure. Unlike previous studies, we model the whole temporal correlation in the data to learn the MI in its general form and use a neural network to estimate the MI-based reward signal to guide the PCMU learning process. This approach is combined with a model-free DRL algorithm known as the Deep Double Q-Learning (DDQL) method. The performance of the complete DDQL-MI algorithm is assessed empirically using an actual SMs dataset and compared with simpler privacy measures. Our results show significant improvements over state-of-the-art privacy-aware demand shaping methods.

eess.SP↗