ABSTRACT The rapid advancement of cloud computing and big data has led to an increasing number of data owners seeking to outsource encrypted data to cloud servers. Searchable encryption (SE) is recognized as a crucial cryptographic primitive for data owners to retrieve data in this context. However, most current SE schemes suffer from low keyword search accuracy and an inability to verify search results. Additionally, encrypted data should be shared with specific data users without compromising their privacy. Therefore, this article introduces a novel fuzzy keyword‐enabled ranked searchable ciphertext‐policy attribute‐based encryption (FKRSCPABE) scheme to address these challenges. First, we combine ciphertext‐policy attribute‐based encryption (CP‐ABE) and public‐key encryption with keyword search (PEKS) to realize fine‐grained access control and efficiently search outsourced encrypted data for data owners. Second, since the document almost certainly contains typographical errors, we implement fuzzy keyword search to ensure meaningful and accurate results. Additionally, we employ probabilistic trapdoors to resist distinguishability attacks. Furthermore, we rank documents using weighted regional scores to improve search accuracy. As a result, our scheme achieves IND‐CCA security by using the classic Canetti transformation. In brief, our scheme is able to achieve fuzzy keyword search and better ranking of search results, while also ensuring data security.

Backdoor attacks involve the injection of a limited quantity of poisoned samples containing triggers into the training dataset. During the inference stage, backdoor attacks can uphold a high level of accuracy for normal examples, yet when presented with trigger-containing instances, the model may erroneously predict them as the targeted class designated by the attacker. This paper addresses the challenge of backdoor attacks by developing a novel method for filtering poisoned samples. We primarily leverage two key characteristics of backdoor attacks: 1) Multiple backdoors can exist simultaneously within a single model; 2) The discovery through Composite Backdoor Attack (CBA) that altering two triggers in a sample to new target labels does not compromise the original functionality of the triggers, yet enables the prediction of the data as a new target class when both triggers are present simultaneously. Therefore, a novel three-stage poisoning data filtering approach, known as Composite Backdoor Poisoning Filtering (CBPF), is proposed as an effective solution. Firstly, utilizing the identified distinctions in output between poisoned and clean samples, a subset of data is partitioned to include both poisoned and clean data. Subsequently, benign triggers are incorporated and labels are adjusted to create new target and benign target classes, thereby prompting the poisoned and clean data to be classified as distinct entities during the inference stage. The experimental results indicate that CBPF is successful in filtering out poisoned data produced by seven advanced attacks on CIFAR-10, GTSRB and ImageNet-12. On average, CBPF attains a notable filtering success rate of 99.88% for these attacks on CIFAR-10. Additionally, the model trained on the uncontaminated samples exhibits sustained high accuracy levels.

At present, cloud computing provides flexible and cost-effective solutions for sharing medical data, particularly electronic health records (EHRs), which are widely used by resource-limited healthcare institutions. However, the sensitive nature of patient data demands strong encryption and privacy protection. Additionally, ensuring data traceability-tracking and monitoring data sources and usage remains a crucial yet unresolved challenge. Current solutions prioritize data integrity during sharing but lack transparency, traceability, and comprehensive integrity audits for cloud-stored medical data. To overcome these limitations, this paper puts forward a blockchain-based framework for medical data sharing, enabling traceable data usage and integrating automatic audit mechanism. Our scheme adopts smart contracts on the blockchain to verify search results, audit cloud data integrity, and allocate service fees equitably based on audits. The proposed solution ensures the integrity of data during the sharing process and storage, while fostering financial fairness between users and cloud service providers. Rigorous experiments demonstrate that our scheme substantially improves verification efficiency, achieving up to a 95% reduction in smart contract gas consumption and an 80%–85% decrease in post-update verification latency compared to existing state-of-the-art methods, making it highly suitable for medical data-sharing environments characterized by frequent updates and stringent data integrity requirements.

Currently, deep neural networks (DNNs) are susceptible to adversarial attacks, particularly when the network's structure and parameters are known, while most of the existing attacks do not perform satisfactorily in the presence of black-box settings. In this context, model augmentation is considered to be effective to improve the success rates of black-box attacks on adversarial examples. However, the existing model augmentation methods tend to rely on a single transformation, which limits the diversity of augmented model collections and thus affects the transferability of adversarial examples. In this paper, we first propose the random diversity ensemble method (RDE-MI-FGSM) to effectively enhance the diversity of the augmented model collection, thereby improving the transferability of the generated adversarial examples. Afterwards, we put forward the random diversity variance ensemble method (RDE-VRA-MI-FGSM), which adopts variance reduction augmentation (VRA) to improve the gradient variance of the enhanced model set and avoid falling into a poor local optimum, so as to further improve the transferability of adversarial examples. Furthermore, experimental results demonstrate that our approaches are compatible with many existing transfer-based attacks and can effectively improve the transferability of gradient-based adversarial attacks on the ImageNet dataset. Also, our proposals have achieved higher attack success rates even if the target model adopts advanced defenses. Specifically, we have achieved an average attack success rate of 91.4% on the defense model, which is higher than other baseline approaches.

As an important cryptographic primitive, searchable encryption (SE) plays a crucial role in performing keyword searching on encrypted texts. However, in order to realize fuzzy keyword search, most fuzzy search encryption schemes utilize wildcards and gram technology to construct fuzzy sets, which consumes a lot of storage and computational resources. Therefore, in this paper, we propose a new verifiable Chinese multi-keyword fuzzy rank searchable encryption (VCMKFRSE) scheme. Firstly, we take advantage of the Yongzi Ba method to convert Chinese keywords into stroke strings, and employ chinese keywords vector generation algorithm to convert the stroke string into a keyword vector. Secondly, we utilize inverted index tables to establish the relationship between keywords and documents. In particular, the relevance score between keywords and documents is calculated by using the three-factor algorithm. Also, we adopt the MinHash function to construct a fuzzy index table for each keyword vector. Thirdly, in order to realize the authentication of search results and avoid receiving useless search results, we build an authentication tag table by using an authentication tag generation function. Afterwards, we apply probabilistic trapdoors to resist distinguishability attacks. At last, our scheme achieves IND-CCA secure and is more efficient comparing with the state of the art. Overall, our proposal achieves fuzzy multi-keyword search and more accurate search result ranking, while ensuring data security and higher efficiency.

The rapid advancement and innovation of the Internet have brought about significant changes in the sharing of data through cloud storage services. Despite these advancements, cloud storage services continue to encounter challenges related to privacy protection and fine-grained access control. To address these issues, we introduce a novel secure attribute-based blockchain (BC) data access control framework (ABBDAC) for real-time attribute tokens in the cloud, enabling collaborative attribute management by multiple authorities. ABBDAC utilizes smart contracts to create encryption policies, attribute tokens for each attribute center, and render policy decisions, thereby diminishing the communication and computation burdens on data users. Furthermore, BC technology aids in securely documenting access control procedures in an auditable manner. Subsequently, a security analysis of the algorithm is conducted, and the system performance is evaluated. Experimental findings suggest that the proposed scheme is both dependable and easily implementable.

With the development of Lie theory, Lie groups have profound significance in many branches of mathematics and physics . In Lie theory, matrix exponential plays a crucial role between Lie groups and Lie algebras . Meanwhile, as finite analogues of Lie groups, finite groups of Lie type also have wide application scenarios in mathematics and physics owning to their unique mathematical structures . In this context, it is meaningful to explore the potential applications of finite groups of Lie type in cryptography. In this paper, we firstly built the relationship between matrix exponential and discrete logarithmic problem (DLP) in finite groups of Lie type. Afterwards, we proved that the complexity of solving non-abelian factorization (NAF) problem is polynomial with the rank n of the finite group of Lie type. Furthermore, combining with the Algebraic Span, we proposed an efficient algorithm for solving group factorization problem (GFP) in finite groups of Lie type. Therefore, it's still an open problem to devise secure cryptosystems based on Lie theory.

Recent research findings suggest that machine learning models are highly susceptible to backdoor poisoning attacks. Backdoor poisoning attacks can be easily executed and achieve high success rates, as the model exhibits anomalous behavior even if a small quantity of malicious data is incorporated into the training dataset. In conventional backdoor defense technologies, fine-tuning is employed as an invasive method that involves adjusting the parameters of model neurons to eliminate backdoors in the attacked model. Nevertheless, this method poses a challenge as the same neurons are responsible for both the original and backdoor tasks, resulting in a decline in the accuracy of the original task during the fine-tuning process. In order to address this issue, we propose a non-invasive approach known as Anti-Backdoor Model (ABM), which does not involve modifying the parameters of the attacked model. ABM employs an external model to counteract the influence of the backdoor task on the attacked model, thereby achieving a balance between eliminating backdoors and preserving the accuracy of the original task. Specifically, our approach involves initially embedding a controllable backdoor in the dataset and leveraging the strong and weak relationships between backdoors to identify a highly concentrated poisoned dataset. Subsequently, we employ the standard training method to train the attacked model (the teacher model). Finally, we utilize this dataset with low volume to train an external model (the student model) that exclusively focuses on backdoors by means of knowledge distillation to counteract the backdoor task in the attacked model (the teacher model). In the experimental part, we assess the effectiveness of ABM by testing eight mainstream attacks on three standard public datasets. Experimental results reveal that ABM exhibits promising efficacy in eliminating the backdoor task while preserving the accuracy of the original task. Our source codes are open athttps://gitee.com/dugu1076/ABM.git.

Summary As the new generation of power scheme, smart grid is proposed to overcome the shortcomings of traditional systems, such as low efficiency and reliability. In this article, a novel lattice‐based secure primitive for privacy‐preserving smart grid communications is proposed, which has the remarkable characteristics, such as scalable multi‐dimensional fine‐grained power data structure and differential privacy security. First, combining with the lattice‐based data encryption technology, while effectively resisting quantum attacks, the method of simultaneous processing of multi‐dimensional data is innovated. Second, through combining the additive homomorphism of the lattice‐based cryptosystem and the Chinese remainder theorem, the data aggregation mechanism that can directly perform homomorphic operations on compressed ciphertext is constructed. Thanks to the above innovative design ideas, the proposed scheme not only significantly improves the efficiency of data communication and processing, greatly reduces the computational cost of the intermediate entity, but also realizes the data confidentiality and information privacy. Finally, observing the decentralized topology of communication nodes in the typical cyber‐physical system of smart grid, the localized differential privacy technology is leveraged to optimize and balance the utility, security, and efficiency of differential privacy. Extensive performance evaluations are conducted to illustrate that the proposed scheme outperforms the state‐of‐the‐art similar schemes in terms of computation complexity and communication cost.

Summary Currently, the protection of users' location privacy, particularly for moveable users, is a major concern for both academia and business. In order to receive required services, a moveable user needs to constantly disclose his/her location information with an untrusted third party in his/her locations, which raises security and privacy issues. To settle the above problems, in this work, we creatively integrate local differential privacy (LDP) with conditional random field (CRF) to facilitate continuous location sharing among moveable users. Firstly, we advance a novel approach of employing CRF to represent users' mobility. After that, we establish a system to provide continuous location sharing by combining the ‐location set and ‐LDP. Finally, we evaluate the system performance on actual data sets. The experimental results indicate that our technique outperforms the planar isotropic mechanism (PIM) and AGENT.

Abstract With the development of Lie theory, Lie groups have attained profound significance in several branches of Mathematics and Physics. In Lie theory, the matrix exponential plays a crucial role between Lie groups and Lie algebras. Meanwhile, as the finite analogue of Lie groups, finite groups of Lie type have potential applications in cryptography due to their unique mathematical structures. In this paper, we first put forward a novel idea of designing cryptosystems based on Lie theory. First of all, combing with discrete logarithm problem and group factorization problem, we proposed several new intractable assumptions based on the matrix exponential in finite groups of Lie type. Subsequently, in analog with Boyen’s scheme (Asiacrypt 2007), we designed a public-key encryption scheme based on the non-abelian factorization problem in finite groups of Lie type. Finally, our proposal was proved to be indistinguishable against adaptively chosen-ciphertext attack in the random oracle model. It is encouraging that our scheme also has the potential to resist against Shor’s quantum algorithm attack.