In recent years, there has been a significant increase in the volume of data and the number of datasets in the healthcare industry, and this trend is expected to continue and intensify. Various strategies are being developed to analyse the data. Nevertheless, these strategies are extensively segregated according to the specific data formats and disorders. Privacy-preserving Hybrid Learning Framework for Healthcare. The framework introduces an hybrid learning technique in order to achieve efficient decision-making. To tackle the challenge of interoperability and heterogeneity with multiple data sources, we propose to integrate a data meshing approach. Furthermore, this paper identifies the potential privacy challenges for machine learning-based healthcare applications that operate with multiple data sources and demand the excessive computation of cloud computing. In addition, we present a comprehensive use case for forecasting cardiovascular disease. The detailed use case and scenarios highlight how our proposal can improve the decision-making process.

With the increasing development of machine learning models in daily businesses, a strong need for intellectual property protection arised. For this purpose, current works suggest to leverage backdoor techniques to embed a watermark into the model, by overfitting to a set of particularly crafted and secret input-output pairs called triggers. By sending verification queries containing triggers, the model owner can analyse the behavior of any suspect model on the queries to claim its ownership. However, when it comes to scenarios where frequent monitoring is needed, the computational overhead of these verification queries in terms of volume demonstrates that backdoor-based watermarking appears to be too sensitive to outlier detection attacks and cannot guarantee the secrecy of the triggers.

Clustering is an unsupervised machine learning technique that outputs clusters containing similar data items. In this work, we investigate privacy-preserving density-based clustering which is, for example, used in financial analytics and medical diagnosis. When (multiple) data owners collaborate or outsource the computation, privacy concerns arise. To address this problem, we design, implement, and evaluate the first practical and fully private density-based clustering scheme based on secure two-party computation. Our protocol privately executes the DBSCAN algorithm without disclosing any information (including the number and size of clusters). It can be used for private clustering between two parties as well as for private outsourcing of an arbitrary number of data owners to two non-colluding servers. Our implementation of the DBSCAN algorithm privately clusters data sets with 400 elements in 7 minutes on commodity hardware. Thereby, it flexibly determines the number of required clusters and is insensitive to outliers, while being only factor 19x slower than today's fastest private K-means protocol (Mohassel et al., PETS'20) which can only be used for specific data sets. We then show how to transfer our newly designed protocol to related clustering algorithms by introducing a private approximation of the TRACLUS algorithm for trajectory clustering which has interesting real-world applications like financial time series forecasts and the investigation of the spread of a disease like COVID-19.

Since machine learning models have become a valuable asset for companies, watermarking techniques have been developed to protect the intellectual property of these models and prevent model theft. We observe that current watermarking frameworks solely target image classification tasks, neglecting a considerable part of machine learning techniques. In this paper, we propose to address this lack and study the watermarking process of various machine learning techniques such as machine translation, regression, binary image classification and reinforcement learning models. We adapt current definitions to each specific technique and we evaluate the main characteristics of the watermarking process, in particular the robustness of the models against a rational adversary. We show that watermarking models beyond classification is possible while preserving their overall performance. We further investigate various attacks and discuss the importance of the performance metric in the verification process and its impact on the success of the adversary.

International audience

Distributed Denial of Service (DDoS) attacks have become one of the most significant problems that affects the user satisfaction by degrading the availability of on-line services. Although intrusion detection systems provide effective mechanism for discriminating various DDoS attacks, they become impotent of detection when bogus packets similar to normal ones are dispatched by the attacker. One idea is to model the normal behavior of the network traffic using time series representation of that traffic together with advanced statistical analysis techniques such as frequency domain analysis for detecting the occurrence frequency (energy) of each basic element in time series. However, frequency domain analysis may become inadequate if the original frequency features are used for the detection anomalies. Therefore, in this work, we propose a hybrid approach that employs frequency domain analysis with sparse representation model to find discriminative characteristics for anomaly-based DDoS detection. The proposed algorithm distinguish abnormal traffic from the normal one based on the energy of time series for the number of packets feature, which is extracted from the time series data by using the sparse representation model. Experimental results show that performance of the proposed algorithm provides better DDoS detection results than the state-of-the-art time-series based approaches in the literature.

research-article Share on Analytical Models for the Scalability of Dynamic Group-key Agreement Protocols and Secure File Sharing Systems Authors: Gokcan Cantali Dept. of Computer Engineering, Bogazici University, Istanbul, Turkey Dept. of Computer Engineering, Bogazici University, Istanbul, TurkeyView Profile , Orhan Ermis Dept. of Computer Engineering, Bogazici University and EURECOM, Sophia-Antipolis, France, Dept. of Computer Engineering, Bogazici University and EURECOM, Sophia-Antipolis, France,View Profile , Mehmet Ufuk Çağlayan Dept. of Computer Engineering, Yasar University, Izmir, Turkey Dept. of Computer Engineering, Yasar University, Izmir, TurkeyView Profile , Cem Ersoy Dept. of Computer Engineering, Bogazici University, Istanbul, Turkey Dept. of Computer Engineering, Bogazici University, Istanbul, TurkeyView Profile Authors Info & Claims ACM Transactions on Privacy and SecurityVolume 22Issue 4November 2019 Article No.: 20pp 1–36https://doi.org/10.1145/3342998Published:25 September 2019Publication History 0citation349DownloadsMetricsTotal Citations0Total Downloads349Last 12 Months22Last 6 weeks1 Get Citation AlertsNew Citation Alert added!This alert has been successfully added and will be sent to:You will be notified whenever a record that you have chosen has been cited.To manage your alert preferences, click on the button below.Manage my AlertsNew Citation Alert!Please log in to your account Save to BinderSave to BinderCreate a New BinderNameCancelCreateExport CitationPublisher SiteGet Access

Information-centric networking (ICN) is a Future Internet paradigm which uses named information (data objects) instead of host-based end-to-end communications. In-network caching is a key pillar of ICN. Basically, data objects are cached in ICN routers and retrieved from these network elements upon availability when they are requested. It is a particularly promising networking approach due to the expected benefits of data dissemination efficiency, reduced delay and improved robustness for challenging communication scenarios in IoT domain. From the security perspective, ICN concentrates on securing data objects instead of ensuring the security of end-to-end communication link. However, it inherently involves the security challenge of access control for content. Thus, an efficient access control mechanism is crucial to provide secure information dissemination. In this work, we investigate Attribute Based Encryption (ABE) as an access control apparatus for information centric IoT. Moreover, we elaborate on how such a system performs for different parameter settings such as different numbers of attributes and file sizes.

With the pervasive use of communications technologies, security of multiparty communication systems becomes crucial more than ever. However, providing a secure communication in distributed and dynamic networks is a challenging issue. Dynamic group key agreement protocols are one of the best candidates to overcome this issue. In dynamic group key agreement protocols, each participant in a group involves into the key computation. Moreover, dynamic group key agreement protocols provide auxiliary dynamic group operations for updating the group key when the set of participants is updated. In this paper, a comparative study on the scalability of dynamic group key agreement protocols is presented to show the best possible group key agreement protocol for specific-sized networks. Furthermore, we present simulations for scalability analysis of dynamic group key agreement protocols. Finally, we analyze and compare the performance of protocols regarding computational and communications costs.

Abstract The pervasive usage of the Internet has made secure group communications a significant issue. Conference‐key agreement protocols provide secure group communications with lower computational cost. Providing key agreements and updates of dynamic groups in an efficient manner is a significant challenge for conference‐key agreement protocols. Auxiliary key agreement operations are needed to solve the challenge. In this paper, we propose an improved conference‐key agreement protocol, called Dynamic Conference‐Key Agreement Protocol, that consists of Initial Conference‐Key Agreement Protocol and Auxiliary Conference‐Key Agreement operations. Dynamic Conference‐Key Agreement Protocol has operations to handle dynamic groups. The proposed protocol has better fault correction and provides the same security level with the existing ones. Copyright © 2014 John Wiley & Sons, Ltd.

The pervasive usage of the Internet has made secure group communications a significant issue. Conference key agreement protocols provide secure group communications against some attacks with lower computational cost in the Internet. However, forward secrecy is a challenging issue in the existing protocols, where it is preserved either the long-term key of a participant is compromised. In this study, we propose an improved conference key agreement protocol with forward secrecy. Besides providing forward secrecy, the proposed protocol preserves the same security level with existing ones.