Deep learning in audio signal processing, such as human voice audio signal classification, is a rich application area of machine learning. Legitimate use cases include voice authentication, gunfire detection, and emotion recognition. While there are clear advantages to automated human speech classification, application developers can gain knowledge beyond the professed scope from unprotected audio signal processing. In this paper we propose the first privacy-preserving solution for deep learning-based audio classification that is provably secure. Our approach, which is based on Secure Multiparty Computation, allows to classify a speech signal of one party (Alice) with a deep neural network of another party (Bob) without Bob ever seeing Alice's speech signal in an unencrypted manner. As threat models, we consider both passive security, i.e. with semi-honest parties who follow the instructions of the cryptographic protocols, as well as active security, i.e. with malicious parties who deviate from the protocols. We evaluate the efficiency-security-accuracy trade-off of the proposed solution in a use case for privacy-preserving emotion detection from speech with a convolutional neural network. In the semi-honest case we can classify a speech signal in under 0.3 sec; in the malicious case it takes $\sim$1.6 sec. In both cases there is no leakage of information, and we achieve classification accuracies that are the same as when computations are done on unencrypted data.
Centralized systems in the Internet of Things---be it local middleware or cloud-based services---fail to fundamentally address privacy of the collected data. We propose an architecture featuring secure multiparty computation at its core in order to realize data processing systems which already incorporate support for privacy protection in the architecture.
The oblivious transfer primitive is sufficient to implement secure multiparty computation. However, secure multiparty computation based only on classical cryptography is severely limited by the security and efficiency of the oblivious transfer implementation. We present a method to efficiently and securely generate and distribute oblivious keys by exchanging qubits and by performing commitments using classical hash functions. With the presented hybrid approach, quantum and classical, we obtain a practical and high-speed oblivious transfer protocol, secure even against quantum computer attacks.…
The Ohio State University（俄亥俄州立大学）；University of California, San Diego（加州大学圣地亚哥分校）
Learning a classifier from private data collected by multiple parties is an important problem that has many potential applications. How can we build an accurate and differentially private global classifier by combining locally-trained classifiers from different parties, without access to any party's private data? We propose to transfer the `knowledge' of the local classifier ensemble by first creating labeled data from auxiliary unlabeled data, and then train a global $ε$-differentially private classifier. We show that majority voting is too sensitive and therefore propose a new risk weighted …
We introduce a framework for navigating through cluttered environments by connecting multiple cameras together while simultaneously preserving privacy. Occlusions and obstacles in large environments are often challenging situations for navigation agents because the environment is not fully observable from a single camera view. Given multiple camera views of an environment, our approach learns to produce a multiview scene representation that can only be used for navigation, provably preventing one party from inferring anything beyond the output task. On a new navigation dataset that we will pub…
We present here a generalization of the work done by Rabin and Ben-Or. We give a protocol for multiparty computation which tolerates any Q^2 active adversary structure based on the existence of a broadcast channel, secure communication between each pair of participants, and a monotone span program with multiplication tolerating the structure. The secrecy achieved is unconditional although we allow an exponentially small probability of error. This is possible due to a protocol for computing the product of two values already shared by means of a homomorphic commitment scheme which appeared origi…
Federated learning (FL) is a privacy-aware data mining strategy keeping the private data on the owners' machine and thereby confidential. The clients compute local models and send them to an aggregator which computes a global model. In hybrid FL, the local parameters are additionally masked using secure aggregation, such that only the global aggregated statistics become available in clear text, not the client specific updates. Federated QR decomposition has not been studied extensively in the context of cross-silo federated learning. In this article, we investigate the suitability of three QR …
University of Campinas（坎皮纳斯大学）；University of Washington Tacoma（华盛顿大学塔科马分校）；Monash University（莫纳什大学）；Aarhus University（奥胡斯大学）
We propose a privacy-preserving Naive Bayes classifier and apply it to the problem of private text classification. In this setting, a party (Alice) holds a text message, while another party (Bob) holds a classifier. At the end of the protocol, Alice will only learn the result of the classifier applied to her text input and Bob learns nothing. Our solution is based on Secure Multiparty Computation (SMC). Our Rust implementation provides a fast and secure solution for the classification of unstructured text. Applying our solution to the case of spam detection (the solution is generic, and can be…
Training Differentially Private Models with Secure Multiparty Computation
School of Engineering and Technology, University of Washington Tacoma（华盛顿大学塔科马分校工程与技术学院）；University of Brasilia（巴西利亚大学）；Monash University（莫纳什大学）；Ghent University（根特大学）；Mila – Quebec AI Institute（米拉-魁北克人工智能研究所）
In recent years, secure multiparty computation (SMC) advanced from a theoretical technique to a practically applicable technology. Several frameworks were proposed of which some are still actively developed. We perform a first comprehensive study of performance characteristics of SMC protocols using a promising implementation based on secret sharing, a common and state-of-the-art foundation. Therefor, we analyze its scalability with respect to environmental parameters as the number of peers, network properties -- namely transmission rate, packet loss, network latency -- and parallelization of …
University of Washington（华盛顿大学）；Monash University（莫纳什大学）；Ghent University（根特大学）
Existing work on privacy-preserving machine learning with Secure Multiparty Computation (MPC) is almost exclusively focused on model training and on inference with trained models, thereby overlooking the important data pre-processing stage. In this work, we propose the first MPC based protocol for private feature selection based on the filter method, which is independent of model training, and can be used in combination with any MPC protocol to rank features. We propose an efficient feature scoring protocol based on Gini impurity to this end. To demonstrate the feasibility of our approach for …
In secure multi-party computation $n$ parties jointly evaluate an $n$-variate function $f$ in the presence of an adversary which can corrupt up till $t$ parties. Almost all the works that have appeared in the literature so far assume the presence of authenticated channels between the parties. This assumption is far from realistic. Two directions of research have been borne from relaxing this (strong) assumption: (a) The adversary is virtually omnipotent and can control all the communication channels in the network, (b) Only a partially connected topology of authenticated channels is guaranteed…
We present initial results on a comprehensive model of structured communications, in which self- adaptation and security concerns are jointly addressed. More specifically, we propose a model of self-adaptive, multiparty communications with secure information flow guarantees. In this model, security violations occur when processes attempt to read or write messages of inappropriate security levels within directed exchanges. Such violations trigger adaptation mechanisms that prevent the violations to occur and/or to propagate their effect in the choreography. Our model is equipped with local and …
Microsoft（微软）；University of Washington Tacoma（华盛顿大学塔科马分校）；Universidade de Brasilia（巴西利亚大学）；Ghent University（根特大学）
Legal and ethical restrictions on accessing relevant data inhibit data science research in critical domains such as health, finance, and education. Synthetic data generation algorithms with privacy guarantees are emerging as a paradigm to break this data logjam. Existing approaches, however, assume that the data holders supply their raw data to a trusted curator, who uses it as fuel for synthetic data generation. This severely limits the applicability, as much of the valuable data in the world is locked up in silos, controlled by entities who cannot show their data to each other or a central a…
This study explores a new security problem existing in various state-of-the-art quantum private comparison (QPC) protocols, where a malicious third-party (TP) announces fake comparison (or intermediate) results. In this case, the participants could eventually be led to a wrong direction and the QPC will become fraudulent. In order to resolve this problem, a new level of trustworthiness for TP is defined and a new QPC protocol is proposed, where a second TP is introduced to monitor the first one. Once a TP announces a fake comparison (or intermediate) result, participants can detect the fraud i…
We propose a secure voting protocol for score-based voting rules, where independent talliers perform the tallying procedure. The protocol outputs the winning candidate(s) while preserving the privacy of the voters and the secrecy of the ballots. It offers perfect secrecy, in the sense that apart from the desired output, all other information -- the ballots, intermediate values, and the final scores received by each of the candidates -- is not disclosed to any party, including the talliers. Such perfect secrecy may increase the voters' confidence and, consequently, encourage them to vote accord…
With the rapid development of various emerging technologies such as the Industrial Internet of Things (IIoT), there is a need to secure communications between such devices. Communication system delays are one of the factors that adversely affect the performance of an authentication system. 5G networks enable greater data throughput and lower latency, which presents new opportunities for the secure authentication of business transactions between IIoT devices. We evaluate an approach to developing a flexible and secure model for authenticating IIoT components in dynamic 5G environments.
A protocol for computing a functionality is secure if an adversary in this protocol cannot cause more harm than in an ideal computation where parties give their inputs to a trusted party which returns the output of the functionality to all parties. In particular, in the ideal model such computation is fair -- all parties get the output. Cleve (STOC 1986) proved that, in general, fairness is not possible without an honest majority. To overcome this impossibility, Gordon and Katz (Eurocrypt 2010) suggested a relaxed definition -- 1/p-secure computation -- which guarantees partial fairness. For t…
FMPC is a novel multiparty computation protocol of arithmetic circuits based on secret-sharing, capable of computing multiplication of secrets with no online communication; it thus enjoys constant online communication latency in the size of the circuit. FMPC is based on the application of Fourier series to Parseval's identity, and introduces the first generalization of Parseval's identity for Fourier series applicable to an arbitrary number of inputs. FMPC operates in a setting where users wish to compute a function over some secret inputs by submitting the computation to a set of nodes, but i…
We describe an asynchronous algorithm to solve secure multiparty computation (MPC) over n players, when strictly less than a 1/8 fraction of the players are controlled by a static adversary. For any function f over a field that can be computed by a circuit with m gates, our algorithm requires each player to send a number of field elements and perform an amount of computation that is O (m/n + \sqrt{n}). This significantly improves over traditional algorithms, which require each player to both send a number of messages and perform computation that is Ω(nm). Additionally, we define the threshold …
Secure multiparty computation (SMC) is a promising technology for privacy-preserving collaborative computation. In the last years several feasibility studies have shown its practical applicability in different fields. However, it is recognized that administration and management overhead of SMC solutions are still a problem. A vital next step is the incorporation of SMC in the emerging fields of the Internet of Things and (smart) dynamic environments. In these settings, the properties of these contexts make utilization of SMC even more challenging since some of its vital premises regarding envi…
Sharing and working on sensitive data in distributed settings from healthcare to finance is a major challenge due to security and privacy concerns. Secure multiparty computation (SMC) is a viable panacea for this, allowing distributed parties to make computations while the parties learn nothing about their data, but the final result. Although SMC is instrumental in such distributed settings, it does not provide any guarantees not to leak any information about individuals to adversaries. Differential privacy (DP) can be utilized to address this; however, achieving SMC with DP is not a trivial t…
The fairness of a secure multi-party quantum key agreement (MQKA) protocol requires that all involved parties are entirely peer entities and can equally influence the outcome of the protocol to establish a shared key wherein no one can decide the shared key alone. However, it is found that parts of the existing MQKA protocols are sensitive to collusion attacks, i.e., some of the dishonest participants can collaborate to predetermine the final key without being detected. In this paper, a multi-party QKA protocol resisting collusion attacks is proposed. Different from previous QKA protocol resis…
The precise design of a system may be considered a trade secret which should be protected, whilst at the same time component manufacturers are sometimes reluctant to release full test data (perhaps only providing mean time to failure data). In this situation it seems impractical to both produce an accurate reliability assessment and satisfy all parties' privacy requirements. However, we present recent developments in cryptography which, when combined with the recently developed survival signature in reliability theory, allows almost total privacy to be maintained in a cryptographically strong …
In the Internet of Things and smart environments data, collected from distributed sensors, is typically stored and processed by a central middleware. This allows applications to query the data they need for providing further services. However, centralization of data causes several privacy threats: The middleware becomes a third party which has to be trusted, linkage and correlation of data from different context becomes possible and data subject lose control over their data. Hence, other approaches than centralized processing should be considered. Here, Secure Multiparty Computation is a promi…
Privacy-preserving data mining has become an important topic. People have built several multi-party-computation (MPC)-based frameworks to provide theoretically guaranteed privacy, the poor performance of real-world algorithms have always been a challenge. Using Principal Component Analysis (PCA) as an example, we show that by considering the unique performance characters of the MPC platform, we can design highly effective algorithm-level optimizations, such as replacing expensive operators and batching up. We achieve about 200$\times$ performance boost over existing privacy-preserving PCA algo…
In game theory, a trusted mediator acting on behalf of the players can enable the attainment of correlated equilibria, which may provide better payoffs than those available from the Nash equilibria alone. We explore the approach of replacing the trusted mediator with an unconditionally secure sampling protocol that jointly generates the players' actions. We characterize the joint distributions that can be securely sampled by malicious players via protocols using error-free communication. This class of distributions depends on whether players may speak simultaneously ("cheap talk") or must spea…
With the increasing popularity of the cloud, clients oursource their data to clouds in order to take advantage of unlimited virtualized storage space and the low management cost. Such trend prompts the privately oursourcing computation, called \emph{multiparty cloud computation} (\MCC): Given $k$ clients storing their data in the cloud, how can they perform the joint functionality by contributing their private data as inputs, and making use of cloud's powerful computation capability. Namely, the clients wish to oursource computation to the cloud together with their private data stored in the c…
A major challenge in the study of cryptography is characterizing the necessary and sufficient assumptions required to carry out a given cryptographic task. The focus of this work is the necessity of a broadcast channel for securely computing symmetric functionalities (where all the parties receive the same output) when one third of the parties, or more, might be corrupted. Assuming all parties are connected via a peer-to-peer network, but no broadcast channel (nor a secure setup phase) is available, we prove the following characterization: 1) A symmetric $n$-party functionality can be securely…
How to achieve differential privacy in the distributed setting, where the dataset is distributed among the distrustful parties, is an important problem. We consider in what condition can a protocol inherit the differential privacy property of a function it computes. The heart of the problem is the secure multiparty computation of randomized function. A notion \emph{obliviousness} is introduced, which captures the key security problems when computing a randomized function from a deterministic one in the distributed setting. By this observation, a sufficient and necessary condition about computi…
Anonymous communication networks are important building blocks for online privacy protection. One approach to achieve anonymity is to relay messages through multiple routers, where each router shuffles messages independently. To achieve anonymity, at least one router needs to be honest. In the presence of an adversary that is controlling a subset of the routers unbiased routing is important for guaranteeing anonymity. However, the routing strategy also influenced other factors such as the scalability and the performance of the system. One solution is to use a fixed route for relaying all messa…
Context: Secure Multiparty Computation (MPC) refers to a family of cryptographic techniques where mutually untrusting parties may compute functions of their private inputs while revealing only the function output. Inquiry: It can be hard to program MPCs correctly and efficiently using existing languages and frameworks, especially when they require coordinating disparate computational roles. How can we make this easier? Approach: We present Symphony, a new functional programming language for MPCs among two or more parties. Symphony starts from the single-instruction, multiple-data (SIMD) semant…
In the setting of secure multiparty computation (MPC), a set of mutually distrusting parties wish to jointly compute a function, while guaranteeing the privacy of their inputs and the correctness of the output. An MPC protocol is called \emph{fully secure} if no adversary can prevent the honest parties from obtaining their outputs. A protocol is called \emph{fair} if an adversary can prematurely abort the computation, however, only before learning any new information. We present highly efficient transformations from fair computations to fully secure computations, assuming the fraction of hones…
Broad Institute of MIT and Harvard（麻省理工学院与哈佛大学博德研究所）
We detail distributed algorithms for scalable, secure multiparty linear regression and feature selection at essentially the same speed as plaintext regression. While the core geometric ideas are simple, the recognition of their broad utility when combined is novel. Our scheme opens the door to efficient and secure genome-wide association studies across multiple biobanks.
In this work we compare two recent multiparty computation (MPC) protocols for private summation in terms of performance. Both protocols allow multiple rounds of aggregation from the same set of public keys generated by parties in an initial stage. We instantiate the protocols with a fast elliptic curve and provide an experimental comparison of their performance for different phases of the protocol. Furthermore, we introduce a technique that allows the computational load of both protocols to be reduced at the expense of protection against collusion tolerance. We prove that both protocols remain…
Business analytics processes are often composed from orchestrated, collaborating services, which are consumed by users from multiple cloud systems (in different security realms), which need to be engaged dynamically at runtime. If heterogeneous cloud systems located in different security realms do not have direct authentication relationships, then it is a considerable technical challenge to enable secure collaboration. In order to address this security challenge, a new authentication framework is required to establish trust amongst business analytics service instances and users by distributing…
Decentralized deep learning plays a key role in collaborative model training due to its attractive properties, including tolerating high network latency and less prone to single-point failures. Unfortunately, such a training mode is more vulnerable to data privacy leaks compared to other distributed training frameworks. Existing efforts exclusively use differential privacy as the cornerstone to alleviate the data privacy threat. However, it is still not clear whether differential privacy can provide a satisfactory utility-privacy trade-off for model training, due to its inherent contradictions…
Modern digitally controlled systems require multiparty authentication and authorization to meet the desired security requirement. This paper describes the design and development of NDN-MPS, an automated solution to support multiparty signature signing and verification for NDN-enabled applications. NDN-MPS suggests several changes and extensions to the existing NDN security solutions. First, it introduces a new type of trust schema to support signing and verification for multiple signers under complex policies such as threshold schemes. Second, it extends the NDN signature format to accommodate…
Digital signatures are widely used in electronic communications to secure important tasks such as financial transactions, software updates, and legal contracts. The signature schemes that are in use today are based on public-key cryptography and derive their security from computational assumptions. However, it is possible to construct unconditionally secure signature protocols. In particular, using quantum communication, it is possible to construct signature schemes with security based on fundamental principles of quantum mechanics. Several quantum signature protocols have been proposed, but n…
We present a framework for experimenting with secure multi-party computation directly in TensorFlow. By doing so we benefit from several properties valuable to both researchers and practitioners, including tight integration with ordinary machine learning processes, existing optimizations for distributed computation in TensorFlow, high-level abstractions for expressing complex algorithms and protocols, and an expanded set of familiar tooling. We give an open source implementation of a state-of-the-art protocol and report on concrete benchmarks using typical models from private machine learning.
Data integration and sharing maximally enhance the potential for novel and meaningful discoveries. However, it is a non-trivial task as integrating data from multiple sources can put sensitive information of study participants at risk. To address the privacy concern, we present a distributed Bayesian learning approach via Preconditioned Stochastic Gradient Langevin Dynamics with RMSprop, which combines differential privacy and homomorphic encryption in a harmonious manner while protecting private information. We applied the proposed secure and privacy-preserving distributed Bayesian learning a…
