Safeguarding cross-silo federated learning with local differential privacy

Wang, Chen; Wu, Xinkui; Liu, Gaoyang; Deng, Tianping; Peng, Kai; Wan, Shaohua

doi:10.1016/j.dcan.2021.11.006

Cited by 40 publications

(17 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Considering that blockchain is essentially a distributed data structure, many scholars have studied the privacy issues in distributed systems [8][9][10][11]. Wu et al [12] propose an efcient identity-based equal test encryption scheme with bilinear pairings to solve the problem of efciently searching encrypted data outsourced to the cloud.…”

Section: Introductionmentioning

confidence: 99%

Privacy-Oriented Transaction for Public Blockchain via Secret Sharing

Huang

et al. 2022

Security and Communication Networks

View full text Add to dashboard Cite

Immutability and traceability are the main reasons for the popularity of blockchain. Nevertheless, its transparency also makes the transactions visible to all participants, which seriously violates the privacy of some dealers. In order to transfer accounts over blockchain, all verifiers should be empowered with the ability to confirm the transactions, leading to the conflict between extensive consensus and individual privacy. Orienting to privacy issues of UTXO (Unspent Transaction Output), this paper exploited the unconditional security of Shamir’s secret sharing to construct a logic-physical map for public chain, which enabled noninteractive transaction verification without privacy infringement. A comprehensive analysis indicated that the proposed scheme is secure under UC (Universal Composability) framework with practical computation and communication overheads.

show abstract

Section: Introductionmentioning

confidence: 99%

Privacy-Oriented Transaction for Public Blockchain via Secret Sharing

Huang

et al. 2022

Security and Communication Networks

View full text Add to dashboard Cite

show abstract

“…In order to avoid threats and violations of users' rights and interests and personal privacy, it is very important to study effective privacy protection methods. Wang et al [ 3 ] proposed the framework for cross-silo Federated Learning with Local Differential Privacy (LDP) mechanism, which can provide strong data privacy protection while still retaining user data statistics to preserve its high utility. Radoglou-Grammatikis et al [ 4 ] applied machine learning and reinforcement learning to the modeling of intrusion detection and prevention system, which effectively improved the detection accuracy.…”

Section: Related Workmentioning

confidence: 99%

Personality Privacy Protection Method of Social Users Based on Generative Adversarial Networks

Sui

Wang

Zheng

et al. 2022

Computational Intelligence and Neuroscience

View full text Add to dashboard Cite

Obscuring or otherwise minimizing the release of personality information from potential victims of social engineering attacks effectively interferes with an attacker’s personality analysis and reduces the success rate of social engineering attacks. We propose a text transformation method named PerTransGAN using generative adversarial networks (GANs) to protect the personality privacy hidden in text data. Making use of reinforcement learning, we use the output of the discriminator as a reward signal to guide the training of the generator. Moreover, the model extracts text features from the discriminator network as additional semantic guidance signals. And the loss function of the generator adds a penalty item to reduce the weight of words that contribute more to personality information in the real text so as to hide the user’s personality privacy. In addition, the semantic and personality modules are designed to calculate the semantic similarity and personality distribution distance between the real text and the generated text as a part of the objective function. Experiments show that the self-attention module and semantic module in the generator improved the content retention of the text by 0.11 compared with the baseline model and obtained the highest BLEU score. In addition, with the addition of penalty item and personality module, compared with the classification accuracy of the original data, the accuracy of the generated text in the personality classifier decreased by 20%. PerTransGAN model preserves users’ personality privacy as found in user data by transforming the text and preserving semantic similarity while blocking privacy theft by attackers.

show abstract

“…It can successfully defend against switching attacks (CCS 2007, Boldyreva et al [8]) and reordering attacks (ISPEC 2007, Shao [9]). In recent years, the combination of signature scheme and blockchain technology [10][11][12], federated learning technology [13], 6G network [14], homomorphic learning [15], network routing protocol [16], edge computing [17], vehicular ad hoc networks [18], and software-defned vehicular network [19][20][21] by applying signature algorithms and encryption algorithms to the experimental scheme to further strengthen the security of the scheme and improve the privacy protection capability of the scheme is also a hot topic. In the blockchain, the digital signature is one of the three basic technologies, and its importance is self-evident.…”

Section: Related Workmentioning

confidence: 99%

New Constructions of Existential Unforgeable Aggregate Signature Scheme from CSP

Zou

Huang

et al. 2022

Security and Communication Networks

View full text Add to dashboard Cite

In future, hundreds of years of mathematical problems that the security of public key cryptography algorithms rely on may be defeated by quantum algorithms. How can a digital signature scheme gracefully balance security and efficiency? This study uses the conjugate search problem and the left self-distributive system to combine and uses the RSA-like algorithm as the underlying structure to propose a new aggregated signature scheme. We, through the EUF game, under the random metaphor model, prove that the security of the scheme satisfies the adaptation unforgeability under selective message attack, the scheme can be finally reduced to the discrete logarithm problem or large prime number decomposition problem. In addition, we can achieve antiquantum attack and exhaustive attack by performing matrix calculations on the message, defining and changing the structure of the matrix by encoding, and setting thresholds for the matrix dimension and the length of the private key. In terms of efficiency, the message signature implementation is linear compared with the expansion rate in terms of storage and computing overhead, and the generation and verification of the final signature pair have nothing to do with the number of users. In addition, the length of the signature is fixed and the size is only the length of a single group, which effectively reduces the generation of public and private key pairs and saves a lot of storage space. The storage space and computational complexity are also effectively improved compared with other solutions.

show abstract

Safeguarding cross-silo federated learning with local differential privacy

Cited by 40 publications

References 25 publications

Privacy-Oriented Transaction for Public Blockchain via Secret Sharing

Privacy-Oriented Transaction for Public Blockchain via Secret Sharing

Personality Privacy Protection Method of Social Users Based on Generative Adversarial Networks

New Constructions of Existential Unforgeable Aggregate Signature Scheme from CSP

Contact Info

Product

Resources

About