检索结果-内蒙古大学图书馆

Multi-Task Faces (Mtf) data Set: A Legally and Ethically Compliant Collection of Face Images for Various Classification Tasks

学校读者我要写书评

暂无评论

SSRN

SSRN 2024年

作者： Haffar, Rami Sánchez, David Domingo-Ferrer, Josep Universitat Rovira i Virgili Department of Computer Engineering and Mathematics CYBERCAT-Center for Cybersecurity Research of Catalonia UNESCO Chair in Data Privacy Av. Països Catalans 26 Catalonia Tarragona43007 Spain

Human facial data hold vast potential for a variety of tasks. However, privacy regulations such as the EU General data Protection Regulation have motivated the removal or limited the publication of facial data sets from the internet due to inadequate data collection methods. While synthetic data sets have been proposed, they do not faithfully represent the real data distribution. Additionally, most available data sets are task-specific. To address these issues, we present the Multi-Task Faces (MTF) data set, a meticulously curated collection of face images labeled for multi-classification tasks -including face, race, gender, and age classification- that strictly follows ethical collection rules and copyright regulations. In this paper, we detail the data collection and processing procedures, and evaluate data quality across five DL models. The reported results constitute a baseline for further research. The MTF data set is accessible on Github. © 2024, The Authors. All rights reserved.

关键词： Image classification

Enhancing Efficiency and data Utility in Longitudinal data Anonymization

学校读者我要写书评

暂无评论

SSRN

SSRN 2023年

作者： Amiri, Fatemeh Sánchez, David Domingo-Ferrer, Josep Universitat Rovira i Virgili Department of Computer Engineering and Mathematics CYBERCAT Center for Cybersecurity Research of Catalonia UNESCO Chair in Data Privacy Av. Paisos Catalans 26 Catalonia TarragonaE-43007 Spain Department of Computer Engineering and Information Technology Hamedan University of Technology Hamedan Iran

Longitudinal data are data collected over time on a set of individuals. The fact that information items are accumulated on each individual over time makes longitudinal data particularly privacy-sensitive. Existing anonymization methods are often ill-suited to ensure privacy-preserving publishing of this kind of data,and current privacy models assume unrealistic attacker knowledge. First, we propose the (k,\beta)^L-privacy model, which assumes the attacker's knowledge to be a subsequence of L quasi-identifiers, and thus provides a more realistic representation of the information actually available to attackers. This also contributes to enhancing the utility of the protected data. Second, we introduce FCLA, an anonymization algorithm to enforce our privacy model that prioritizes data utility while effectively mitigating identity and attribute disclosures, as well as skewness attacks in longitudinal data. FCLA partitions sequences into groups and anonymizes them independently, a task that is straightforward to parallelize. Reported experiments demonstrate that FCLA outperforms existing methods at preserving data utility while satisfying strict privacy constraints. The time complexity analysis and execution time of FCLA demonstrate that FLCA is more efficient, leading to better scalability. © 2023, The Authors. All rights reserved.

关键词： privacy-preserving techniques

Multi-Task Faces (MTF) data Set: A Legally and Ethically Compliant Collection of Face Images for Various Classification Tasks

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Haffar, Rami Sánchez, David Domingo-Ferrer, Josep Universitat Rovira i Virgili Department of Computer Engineering and Mathematics CYBERCAT-Center for Cybersecurity Research of Catalonia UNESCO Data Privacy Av. Països Catalans 26 Catalonia Tarragona43007 Spain

Human facial data hold tremendous potential to address a variety of classification problems, including face recognition, age estimation, gender identification, emotion analysis, and race classification. However, recent privacy regulations, such as the EU General data Protection Regulation and others, have restricted the ways in which human images may be collected and used for research. As a result, several previously published data sets containing human faces have been removed from the internet due to inadequate data collection methods that failed to meet privacy regulations. data sets consisting of synthetic data have been proposed as an alternative, but they fall short of accurately representing the real data distribution. On the other hand, most available data sets are labeled for just a single task, which limits their applicability. To address these issues, we present the Multi-Task Faces (MTF) image data set, a meticulously curated collection of face images designed for various classification tasks, including face recognition, as well as race, gender, and age classification. The MTF data set has been ethically gathered by leveraging publicly available images of celebrities and strictly adhering to copyright regulations. In this paper, we present this data set and provide detailed descriptions of the followed data collection and processing procedures. Furthermore, we evaluate the performance of five deep learning (DL) models on the MTF data set across the aforementioned classification tasks. Additionally, we compare the performance of DL models over the processed MTF data and over raw data crawled from the internet. The reported results constitute a baseline for further research employing these data. The MTF data set can be accessed through the following link (please cite the present paper if you use the data set): https://***/RamiHaf/MTF data set. © 2023, CC BY.

关键词： Image classification

FL-Defender: Combating Targeted Attacks in Federated Learning

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Jebreel, Najeeb Domingo-Ferrer, Josep Universitat Rovira i Virgili Department of Computer Engineering and Mathematics CYBERCAT-Center for Cybersecurity Research of Catalonia Unesco in Data Privacy Av. Països Catalans 26 Catalonia Tarragona43007 Spain

Federated learning (FL) enables learning a global machine learning model from local data distributed among a set of participating workers. This makes it possible i) to train more accurate models due to learning from rich joint training data, and ii) to improve privacy by not sharing the workers' local private data with others. However, the distributed nature of FL makes it vulnerable to targeted poisoning attacks that negatively impact the integrity of the learned model while, unfortunately, being difficult to detect. Existing defenses against those attacks are limited by assumptions on the workers' data distribution, may degrade the global model performance on the main task and/or are ill-suited to high-dimensional models. In this paper, we analyze targeted attacks against FL and find that the neurons in the last layer of a deep learning (DL) model that are related to the attacks exhibit a different behavior from the unrelated neurons, making the last-layer gradients valuable features for attack detection. Accordingly, we propose FL-Defender as a method to combat FL targeted attacks. It consists of i) engineering more robust discriminative features by calculating the worker-wise angle similarity for the workers' last-layer gradients, ii) compressing the resulting similarity vectors using PCA to reduce redundant information, and iii) re-weighting the workers' updates based on their deviation from the centroid of the compressed similarity vectors. Experiments on three data sets with different DL model sizes and data distributions show the effectiveness of our method at defending against label-flipping and backdoor attacks. Compared to several state-of-the-art defenses, FL-Defender achieves the lowest attack success rates, maintains the performance of the global model on the main task and causes minimal computational overhead on the server. Copyright © 2022, The Authors. All rights reserved.

关键词： Deep learning

Enhanced Security and privacy via Fragmented Federated Learning

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Jebreel, Najeeb Moharram Domingo-Ferrer, Josep Blanco-Justicia, Alberto Sánchez, David The Unesco in Data Privacy Cybercat Center for Cybersecurity Research of Catalonia Department of Computer Engineering and Mathematics Universitat Rovira i Virgili Av. Països Catalans 26 Catalonia TarragonaE-43007 Spain

In federated learning (FL), a set of participants share updates computed on their local data with an aggregator server that combines updates into a global model. However, reconciling accuracy with privacy and security is a challenge to FL. On the one hand, good updates sent by honest participants may reveal their private local information, whereas poisoned updates sent by malicious participants may compromise the model's availability and/or integrity. On the other hand, enhancing privacy via update distortion damages accuracy, whereas doing so via update aggregation damages security because it does not allow the server to filter out individual poisoned updates. To tackle the accuracy-privacy-security conflict, we propose fragmented federated learning (FFL), in which participants randomly exchange and mix fragments of their updates before sending them to the server. To achieve privacy, we design a lightweight protocol that allows participants to privately exchange and mix encrypted fragments of their updates so that the server can neither obtain individual updates nor link them to their originators. To achieve security, we design a reputation-based defense tailored for FFL that builds trust in participants and their mixed updates based on the quality of the fragments they exchange and the mixed updates they send. Since the exchanged fragments' parameters keep their original coordinates and attackers can be neutralized, the server can correctly reconstruct a global model from the received mixed updates without accuracy loss. Experiments on four real data sets show that FFL can prevent semi-honest servers from mounting privacy attacks, can effectively counter poisoning attacks and can keep the accuracy of the global model. Copyright © 2022, The Authors. All rights reserved.

关键词：

privacy-Preserving Computation of the Earth Mover’s Distance 23rd

学校读者我要写书评

暂无评论

Privacy-Preserving Computation of the Earth Mover’s Distanc...

23rd International Conference on Information Security, ISC 2020

作者： Blanco-Justicia, Alberto Domingo-Ferrer, Josep Department of Computer Engineering and Mathematics Universitat Rovira i Virgili UNESCO Chair in Data Privacy CYBERCAT-Center for Cybersecurity Research of Catalonia Av. Països Catalans 26 TarragonaCatalonia Spain

ISBN: (纸本)9783030629731

The Wasserstein distance, also known in computer science as the Earth Mover’s Distance (EMD) is a distance metric between two probability distributions. EMD has often been used as a distance metric to compare images and documents, and is central to privacy models such as t-closeness. In this work, we show that, given one-dimensional discrete probability distributions, the computation of EMD can be reduced to the computation of the cardinality of the intersection of two sets. We then use a private matching scheme to create a privacy-preserving computation protocol for EMD: two parties can compute EMD between their privately-owned documents without revealing them to the other party. We demonstrate our proposal by implementing a privacy-preserving reverse image search, where images are kept encrypted at an external server. © 2020, Springer Nature Switzerland AG.

关键词： Probability distributions

Multi-Dimensional Randomized Response

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Domingo-Ferrer, Josep Soria-Comas, Jordi Universitat Rovira i Virgili Dept. of Computer Engineering and Mathematics UNESCO Chair in Data Privacy CYBERCAT-Center for Cybersecurity Research of Catalonia Av. Països Catalans 26 Tarragona Catalonia43007 Spain Catalan Data Protection Authority Rosselló 214 Esc. A 1r 1a Barcelona Catalonia08008 Spain

In our data world, a host of not necessarily trusted controllers gather data on individual subjects. To preserve her privacy and, more generally, her informational self-determination, the individual has to be empowered by giving her agency on her own data. Maximum agency is afforded by local anonymization, that allows each individual to anonymize her own data before handing them to the data controller. Randomized response (RR) is a local anonymization approach able to yield multi-dimensional full sets of anonymized microdata that are valid for exploratory analysis and machine learning. This is so because an unbiased estimate of the distribution of the true data of individuals can be obtained from their pooled randomized data. Furthermore, RR offers rigorous privacy guarantees. The main weakness of RR is the curse of dimensionality when applied to several attributes: as the number of attributes grows, the accuracy of the estimated true data distribution quickly degrades. We propose several complementary approaches to mitigate the dimensionality problem. First, we present two basic protocols, separate RR on each attribute and joint RR for all attributes, and discuss their limitations. Then we introduce an algorithm to form clusters of attributes so that attributes in different clusters can be viewed as independent and joint RR can be performed within each cluster. After that, we introduce an adjustment algorithm for the randomized data set that repairs some of the accuracy loss due to assuming independence between attributes when using RR separately on each attribute or due to assuming independence between clusters in cluster-wise RR. We also present empirical work to illustrate the proposed methods. Copyright © 2020, The Authors. All rights reserved.

关键词： Machine learning

Big data anonymization requirements vs privacy models 15

学校读者我要写书评

暂无评论

Big data anonymization requirements vs privacy models

15th International Joint Conference on e-Business and Telecommunications, ICETE 2018

作者： Domingo-Ferrer, Josep Universitat Rovira i Virgili Department of Computer Science and Mathematics CYBERCAT-Center for Cybersecurity Research of Catalonia UNESCO Chair in Data Privacy Av. Països Catalans 26 Tarragona Catalonia43007 Spain

ISBN: (纸本)9789897583193

The big data explosion opens unprecedented analysis and inference possibilities that may even enable modeling the world and forecasting its evolution with great accuracy. The dark side of such a data bounty is that it complicates the preservation of individual privacy: a substantial part of big data is obtained from the digital track of our activity. We focus here on the privacy of subjects on whom big data are collected. Unless anonymization approaches are found that are suitable for big data, the following extreme positions will become more and more common: nihilists, who claim that privacy is dead in the big data world, and fundamentalists, who want privacy even at the cost of sacrificing big data analysis. In this article we identify requirements that should be satisfied by privacy models to be applicable to big data. We then examine how well the two main privacy models (k-anonymity and Ε-differential privacy) satisfy those requirements. Neither model is entirely satisfactory, although k-anonymity seems more amenable to big data protection. Finally, we highlight connections between the previous two privacy models and other privacy models that might result in synergies between them in order to tackle big data: the principles underlying all those models are deniability and permutation. Future research attempting to adapt the current privacy models for big data and/or design new models will have to adhere to those two underlying principles. As a side result, the above inter-model connections allow gauging what is the actual protection afforded by differential privacy when Ε is not sufficiently small. Copyright © 2018 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved

关键词： Big data

学校读者我要写书评

暂无评论

Co-utility

Progress in Artificial Intelligence 2016年第2期5卷 105-110页

作者： Josep Domingo-Ferrer David Sánchez Jordi Soria-Comas UNESCO Chair in Data Privacy Department of Computer Engineering and Mathematics Universitat Rovira i Virgili Tarragona Catalonia Spain

Peer-to-peer scenarios are increasingly important in globalized information societies. To guide the interaction between (rational) agents in these P2P settings, self-enforcing collaborative protocols are desirable, so that agents have no motivation to deviate from them. Moreover, to increase social welfare and ensure the success of protocols, it would be desirable to make the collaboration between agents mutually beneficial. In this paper, we present the notion of co-utility, which refers to protocols in which mutual help is the best rational option to take, even for purely selfish agents. This concept is illustrated through several P2P scenarios, framed both in the information society and in the physical world, for which we present and discuss co-utile self-enforcing protocols.

关键词：