Multimodal recommender systems have emerged as a pivotal paradigm for harnessing diverse data modalities to deliver personalized services. Contemporary research predominantly focuses on integrating heterogeneous modality information through graph learning. However, these approaches face two key challenges: (1) the inherent complexity of modalities, characterized by entangled redundant signals and noise; and (2) the challenge of effectively integrating multimodal representations, each of which may exert varying degrees of influence on users' preferences. To address these challenges, we propose a novel Collaboration-Guided Multimodal Disentanglement and Hierarchical Fusion for Recommendation (DHMRec), which simultaneously achieves intra-modal denoising disentanglement and inter-modal hierarchical fusion. Specifically, we introduce a collaboration-related modality disentanglement module to distinguish between modality-common and modality-specific features. Then, through multi-view graph learning to capture both item-item dependencies and user-item interaction patterns. Additionally, we implement hierarchical fusion between the disentangled multimodal features and ID embeddings using a positive-negative attention-aware fusion module and an interaction distribution-based alignment module. Extensive experiments on three benchmarks demonstrate that our DHMRec surpasses various state-of-the-art baselines, highlighting its effectiveness in intra-modal disentanglement and multimodal features fusion.

Reinforcement learning (RL) has been increasingly adopted in IoT systems for tasks such as resource allocation and control. However, in privacy-critical and resource-constrained environments, existing privacy-preserving RL schemes suffer from high computational cost, slow convergence, and limited scalability due to the use of homomorphic encryption or differential privacy. We propose a distributed privacy-preserving Q-learning framework that enables secure and efficient policy updates across multiple clients. Each client independently trains a local Q-table and encrypts it using a sparse matrix transformation combined with additive secret sharing of structured perturbations. The encrypted Q-tables are uploaded to a cloud server for aggregation and averaging without decryption. The encrypted global Q-table is then returned and decrypted locally using the inverse sparse matrix. Experimental results on four benchmark environments (CartPole-v1, MountainCar-v0, Acrobot-v1, and LunarLander-v3) show that our scheme achieves up to 92% reduction in computation time compared to the FHE-based method, while maintaining comparable reward performance and faster convergence.

Machine learning algorithm for multi-modal image segmentation is extensively employed in medical analysis and diagnosis. Clustering represents a mainstream approach for image segmentation, with the fuzzy c-means and particle swarm optimization (FCM-PSO) algorithm garnering significant attention. As image segmentation tasks have substantial computational costs, the outsourcing scheme offers an effective solution by leveraging cloud servers to execute complex computations. Given that medical images contain sensitive patient information, the image segmentation outsourcing scheme must ensure data privacy and confidentiality. In this paper, we propose a secure outsourcing scheme for the FCM-PSO based image segmentation algorithm through a novel sparse matrix encryption method. By analyzing each stage of the image segmentation algorithm, we delegate the computationally intensive task of calculating the Euclidean distance to an untrusted cloud server. We utilize sparse matrices to obscure the private image data. These matrices are created by incorporating multiple small-sized random invertible matrices, thereby circumventing the local storage of generation factors. Additionally, we implement a lightweight verification method to verify the correctness of returned results. Experimental results show that our scheme improves the efficiency of the image segmentation task by 29.99% to 49.50% with the increasing of image set, compared to the original algorithm executed locally.

With the rapid development of scientific research, a large amount of literature materials (e.g., published papers) has brought a serious information overload problem to researchers. For those novices who have just stepped into a certain research field, the fact that they do not yet know their own research direction, coupled with the huge amount of literature materials and their varying quality, makes it even more difficult for them to retrieve high-quality papers. To this end, we propose a Cross-community academic Knowledge Graph based approach for machine learning paper Recommendation (CKGR). Considering the hierarchical structure of cross-community knowledge graphs, we utilize knowledge propagation in hyperbolic space for entity representation learning. To alleviate the data sparsity problem as well as to learn better entity representations, we further introduce a preference migration module and contrastive learning. Meanwhile, considering the semantic relationships among entities, we introduce textual information to enhance the connection among interacting nodes for better recommendation tasks.

In this work, we propose a privacy-preserving face recognition scheme based on secure similarity comparison on encrypted data. We innovatively split sensitive face embeddings into two shares and encrypt them to guarantee data privacy. We present a novel matrix blinding method to conduct face similarity computation on encrypted face embeddings. Furthermore, we design an effective re-encryption method to achieve non-local secure data update, which reduces the risk of data leakage. Simulation experiments demonstrate that the proposed scheme completes face recognition tasks securely and efficiently. With different size of datasets, scale of embeddings, and number of queries, our scheme accomplishes face recognition tasks at low costs without obvious accuracy penalty.

With the rapid development of artificial intelligence and big data technologies, gait recognition has become a key method for identity authentication. As a unique biometric characteristic, gait is difficult to counterfeit and supports long-distance, non-contact authentication, making it ideal for security, surveillance, and health monitoring. However, traditional gait authentication faces privacy and efficiency challenges. This paper presents a privacy-preserving gait authentication scheme based on partial Euclidean distance calculation, and the scheme encrypts gait features using block-diagonal orthogonal matrices and random perturbation vectors, enabling efficient encrypted computation on the cloud. Experimental results demonstrate that the proposed scheme improves processing efficiency and matching accuracy while protecting privacy. For instance, on the CASIA-B dataset, it reduces computation time by approximately 30% without compromising accuracy.

Saber, as one of the candidates of the NIST Post Quantum Cryptography (PQC) Standardization, is a post-quantum Key Encapsulation Mechanism (KEM) that provides strong security guarantees against quantum adversaries. Due to the high computational complexity, enhancing its efficiency is essential for real-world deployment. In practical applications, the high computational cost of polynomial multiplication remains a major bottleneck to its performance. To address this, we propose a novel optimization method for Saber by introducing a Fermat number transform (FNT) as a key algorithmic improvement. By leveraging a Fermat prime as one of the moduli, the modular multiplications are efficiently implemented by bitshift operations, thereby reducing the arithmetic complexity in negacyclic convolutions. We integrate FNT with a K-reductionbased modular reduction scheme which eliminates the need for precomputation or multiplications and enables reduction to be completed using only shifts and additions. To fully exploit the Intel hardware architecture, we design a high-performance AVX2-based vectorized implementation by utilizing 256-bit SIMD instructions. Specifically, FNT is optimized by using shift-based butterfly operations tailored to the structure of Fermat primes, while the modular reduction is accelerated by bitwise and shift instructions, ensuring high throughput and memory efficiency. Our experimental evaluation shows that the proposed polynomial multiplication scheme achieves a speedup of 14.7 % and reduces cycle counts by up to 8 % compared to previous state-of-the-art implementations. These improvements highlight the efficacy of our method, making it highly suitable for high-performance cryptographic applications.

In the field of machine learning, PCA based face recognition is widely applied in identity authentication and access control. Due to limited storage and computational capabilities, the client often needs to outsources face recognition tasks to cloud servers, which brings privacy leakage risks. Existing privacy-preserving schemes calculate the inner product of encrypted vectors to perform face matching, but they involve complex operations, making efficient recognition challenging and lacking in automatic verification of computational integrity. In this paper, we propose a blockchain-aided privacy-preserving PCA based face recognition scheme. Our approach uses sparse matrices to construct keys, reducing the number of non-zero elements in matrix operations. The scheme also introduces a blockchainaided verification and payment mechanism based on hash commitments to verify the integrity of computation tasks. This mechanism allows encrypted face images and face recognition results to be uploaded to the blockchain, effectively reducing the number of interactions between the client and the cloud server. Experimental results demonstrate that this scheme reduces the overall execution time by 43% without compromising recognition accuracy, achieving a recognition accuracy rate of 99.35%.

In the Internet of Things (IoT) scenarios, edge computing assists in completing machine learning on resource-constrained terminal devices. As one of the most significant operations, large-scale matrix multiplication remains a huge efficiency bottleneck. Existing distributed computation approaches typically decompose the matrix computation into subtasks of the same scale, overlooking edge computing environments with unbalanced computing resources. In this paper, we propose a secure distributed matrix multiplication outsourcing scheme in edge computing with unbalanced resources. Specifically, the large-scale matrix multiplication is decomposed into several subtasks of varying scales according to unbalanced edge computing resources, which achieves better distributed computational performance. A novel matrix blinding method is presented by employing perturbation matrices and permutation matrices to guarantee input and output privacy. Experimental results demonstrate that our scheme improves the efficiency by 51.13% to 98.79% compared to traditional matrix multiplication without outsourcing. Additionally, our scheme outperforms state-of-the-art outsourcing schemes, with an average improvement of 9.36% in edge computing with balanced resources and 14.83% in unbalanced environments.

Profiled side-channel analysis presents a significant risk to embedded devices in Internet of Things (IoT). Typically, a single trace is insufficient to successfully key recovery in practical scenarios. It still requires several traces based on Bayes’ posterior probability. In this article, we introduce a chosen-plaintext (CP) strategy into the deep learning-based profiled attacks to improve the attack efficiency. First, we present a general strategy to profile the leakage model by exploiting the sensitivity analysis and clustering analysis. The leakage model derived from deep neural network is to characterize the leakage of the target algorithm. Second, we propose an adaptive CP method in the deep learning-based attack, transforming the conditional probability distribution of the leakage into the entropy of the key candidates under the profiled leakage model. Finally, we evaluate the efficiency of the attack by practical measurements. The results demonstrate that the proposed method requires fewer traces to retrieve the key of AES on devices of different types, e.g., Smartcard, FPGA, and ARM. Moreover, our attack improves the attack efficiency on masked implementations.

For the huge amount of data from the Internet of Things (IoT) devices, multiple kernel learning is a widely concerned issue in data analyzing, among which the multiple kernel fuzzy clustering (MKFC) algorithm is an effective approach for extracting linear features in high-dimensional space. For a time-consuming multiple kernel clustering task, it is meaningful to find a secure and efficient outsourcing scheme in the edge-end collaborative architecture, which utilizes edge computing resources while resisting untrusted edge servers. However, existing secure outsourcing schemes cannot align well with the distributed and real-time characteristics of edge computing due to their complex encryption processes. In this article, we propose a secure MKFC outsourcing scheme based on a novel matrix blinding method. The proposed novel matrix blinding method conducts two related encryption operations with disturbance terms, which avoids specific disturbance elimination computations, to reduce the computational burdens in the decryption phase. Additionally, we introduce a sampling verification method to detect the server’s deceptive behaviors. The theoretical analysis demonstrates that our scheme guarantees data privacy and has the capability to verify incorrect results. The experimental results indicate that our scheme is 6.73% superior to other schemes on average when conducting matrix outsourcing computation and enhances the efficiency of conducting the MKFC algorithm by 10.44% to 55.70% on different datasets.

Eigenvalue decomposition (EVD) is a fundamental yet time-consuming operation with extensive applications in Internet of Things (IoT). When the matrix dimension reaches millions, resource-limited IoT devices struggle to perform such computationally expensive operations. Edge computing, with its plentiful computing resources, offers an effective solution to this problem. However, privacy concerns arise because outsourced tasks may contain sensitive user data. In this article, we propose the first privacy-preserving, edge-assisted EVD outsourcing scheme that securely enables users to outsource EVD tasks to edge servers. We design a privacy-preserving matrix transformation method to encode the original data, ensuring that edge servers cannot access users’ private information. Additionally, we design a verification scheme that enables the user to verify the correctness of the results returned by the edge servers. Our protocol supports parallel computation by multiple edge servers, thus enhancing the efficiency of EVD. The feasibility of our proposed scheme is demonstrated through both theoretical and experimental perspectives.

Transformers based on Self-Attention (SA) mechanism have demonstrated unrivaled superiority in numerous areas. Compared to RNN-based networks, Transformers can learn the temporal dependency representation of an entire sequence in parallel, while efficiently dealing with long-range dependencies. However, the $\mathcal {O}(L^{2})$O(L2) ($L$L denotes the length of the sequence) computational complexity of the SA mechanism and the high memory usage make the construction cost of the Transformer-based model prohibitively expensive. To address these challenges, we propose a Transformer-like model, HPformer: Low-Parameter Transformer with Temporal Dependency Hierarchical Propagation. HPformer first chunks the sequence into $K$K ($K = \left\lceil \log {L} \right\rceil + 1$K=logL+1, $\left\lceil \cdot \right\rceil$· denotes ceiling operation) sequence segments, then leverages the hierarchical propagation mechanism with $\mathcal {O}(L)$O(L) computational complexity to learn the temporal dependencies between the segments and within the segments, and ultimately generates $K$K vectors as $Key$Key matrices. This reduces the complexity of the SA mechanism from $\mathcal {O}(L^{2})$O(L2) to $\mathcal {O}(L\log {L})$O(LlogL). In addition, we employ a strategy of sharing $Key$Key and $Value$Value matrices between layers to build the HPformer, thus reducing memory usage. Extensive experiments based on public health informatics benchmark and Long-Range Arena (LRA) benchmark have demonstrated that HPformer has advantages over Transformer-based models in terms of memory usage and efficiency.

This study explores the metric of group closeness centrality within the framework of social networks, a departure from the traditional analysis focused solely on the significance of individual nodes. Given the intricate dynamics observed in networks governed by various stakeholders, we introduce a framework that preserves privacy through the application of a greedy algorithm. This approach is designed to evaluate the collective influence of groups while ensuring the confidentiality of individual data. Furthermore, we employ Oblivious Random Access Memory (ORAM) [1] within cloud servers to conceal access patterns, thereby enhancing data privacy. Through comprehensive experimentation across three real-world social network datasets within the MP-SPDZ framework [2], dedicated to secure multi-party computation, we demonstrate the efficiency of our proposed methods.

The large-scale matrix eigenvalue computation, as a basic mathematical tool, has been widely used in many fields such as face recognition and data analysis. However, local terminal devices lack sufficient resources to undertake heavy computational tasks, which poses a challenge to the applications of eigenvalue computation. In this paper, we propose the first privacy-preserving edge-assisted computation scheme for solving the largest eigenvalue and corresponding eigenvector. We propose a privacy-preserving transformation method to protect data privacy and prevent edge servers from retrieving sensitive information. Mean-while, we design a verification scheme to ensure the correctness of the results returned by the edge servers. In addition, we design a distributed parallel computing scheme to ensure the efficiency of edge computation. Through theoretical analysis and simulation experiments, we verify the feasibility and efficiency of our proposed scheme.

Lattice-based Post-Quantum Cryptography (PQC) can effectively resist the quantum threat to blockchain's underlying cryptographic algorithms. Blockchain node decryption is one of the most commonly used cryptographic computations in blockchain systems, and polynomial multiplication, a time-consuming operation for decryption, is one of the factors limiting blockchain efficiency. This paper proposes a novel distributed computing algorithm for polynomial multiplication, applicable in blockchain decryption. By splitting polynomials into lower-degree terms and delegating tasks to distributed nodes, our approach reduces computation time. A novel verification strategy based on the Karatsuba algorithm ensures result accuracy. The experimental results demonstrate that our proposed scheme improves the execution efficiency of NTT and INTT operations by approximately 47.8% and 52.4%, and reduces Kyber decryption time by up to 23.5%.

Endometrial cancer is one of the most common gynecological malignancies affecting women worldwide, posing a serious threat to women’s health. Moreover, the identification of drug-target interactions (DTIs) is typically a time-consuming and costly critical step in drug discovery. In order to identify potential DTIs to enhance targeted therapy for endometrial cancer, we propose a deep learning model named DTRE (Drug-Target Relationship Enhanced) based on a heterogeneous graph to predict DTIs, which utilizes the relationships between drugs and targets to effectively capture their interactions. In the heterogeneous graph, nodes represent drugs and targets, and edges represent their interactions, then the representations of drugs and targets are learned through graph convolutional network, graph attention network and attention mechanism. Experimental results on the dataset proposed in this paper show that the AUC and AUPR of DTRE achieve 0.870 and 0.872 respectively, significantly outperforming comparative models and indicating that DTRE can effectively predict DTIs when applied to large-scale data. Additionally, DTRE also predicts the potential DTIs for endometrial cancer, providing new insights into targeted therapy for it.

With the advancement of the Internet of Things (IoT), numerous machine learning applications on IoT are encountering performance bottlenecks. Graph embedding is an emerging type of machine learning that has achieved commendable results in areas such as network anomaly detection, malware detection, IoT device management, and service recommendation within the Internet of Things. However, for some resource-constrained IoT devices, computing graph embedding algorithms is highly complex and time-consuming. In this paper, we introduce an efficient and secure distributed outsourcing scheme, employing four non-colluding cloud servers to facilitate the computation of graph embedding for IoT devices. Our scheme utilizes a novel blinding factor generated through QR decomposition to blind matrices containing sensitive information. We partition the blinded matrix into several segments, distributing different small matrix blocks across four servers, each of which executes only a portion of the computational tasks. The proposed outsourcing solution ensures the privacy of input and output information is not compromised. In our scheme, we utilize an effective verification method that can detect the erroneous behaviors of cloud servers with a probability close to one. Theoretical analysis and experimental results indicate that our solution achieves a computational efficiency of (35m2+2m)/(3m3) compared to the original algorithm.

Decision tree classifiers are pervasively applied in a wide range of areas, such as healthcare, credit-risk assessment, spam detection, and many more. To ensure effectiveness and efficiency, clients usually choose to adopt classification services that are offered by model providers. However, the required data interactions in the evaluation process raise privacy concerns for both the provider and the client, indicating an imminent need for private decision tree evaluation (PDTE). Recently, some works, e.g., [1] (ESORICS'19) and [2] (NDSS'21), try to achieve PDTE by secure outsourcing computation. However, to hide the decision tree structure, [1] and [2] require non-complete decision trees to be made complete by padding dummy nodes, which lead to exponential (provider-side and cloud-side) computation and communication complexity in the depth of the decision tree. This is especially impractical for deep but sparse decision trees. In this paper, we propose a secure and efficient outsourced PDTE protocol with a focus on sparse trees. We avoid padding dummy nodes by vector dot products in outsourcing settings. Through experiments, we show the competitive performance of our design. Compared with [2] on Spambase dataset in the cloud-side, we are 486× more communication efficient in offline phase and 15× more communication efficient in online phase.

Approaches predicting the results of mutation testing by machine learning have been proposed to reduce the cost of mutation testing. The predictive approaches based on PIE theory and approaches based on natural language have been proposed. However, both PIE-based and natural language-based approaches have disadvantages, leading to a reduction in effectiveness at the test case level prediction. In order to predict at the test case level and improve the effectiveness of prediction, we propose Natural Language and PIE Predictive Mutation Testing (NLPIE-PMT), which combines advantages of PIE-based and natural language-based approaches and predict whether each test case kills each mutant in the cross-version scenario. The experimental results on subjects in Defects4J show that NLPIE-PMT can predict whether each test case kill each mutant with the average F1-score of 0.811, which is 0.135 and 0.046 higher than the PIE-based baseline and the natural language-based baseline respectively. NLPIE-PMT also performs better than the baselines in predicting mutation score.

An online service system may experience various performance faults during operation. Detecting and locating these faults after they occur can significantly impact the user experience and lead to significant losses. Therefore, it is necessary to predict faults before they occur. Existing methods for fault prediction typically only predict the possibility of fault, without providing more granular predictions, such as the type of fault. This can make troubleshooting more difficult for developers. In this paper, we propose a fine-grained fault prediction method called FSFP, which not only predicts the possibility of fault but also identifies the type of fault that may occur. The method initially collects performance monitoring metrics from the runtime system, including two types: normal operation and abnormal conditions. It then utilizes cross-attention to capture the interdependencies between these two types of monitoring metrics, followed by the construction of a multi-label classification model. We evaluated FSFP by injecting faults into a benchmark microservice system. In terms of predicting the possibility of fault, FSFP achieved a precision of 0.999, a recall of 0.998, and an F1 score of 0.999. In terms of predicting the type of fault, FSFP achieved an exact match ratio of 0.955 and a Hamming loss of 0.017. In terms of predicting six specific types of faults, FSFP achieved four optimal F1 scores.

The purpose of drug recommendation is to predict the effective and safe drug combinations required for the current visit based on the historical medical data of patients. How to better mine the hidden relationship in the medical data and effectively improve the accuracy of drug recommendation are research hotspots in the medical field. This paper proposes a Collaborative Cross-attention Drug Recommendation model (CCDR) based on patient and medical relationship representations, which mines medical data from two aspects to enhance the representation ability of the model. CCDR obtains patient representation vectors by modeling the patients’ historical sequence data using Bidirectional Gated Recurrent Unit. Meanwhile, CCDR designs a medical graph structure data learning method based on relationship division to better capture the complex association relationships among diagnoses, procedures, and drugs. Finally, the representation capability of the model is enhanced by introducing a collaborative cross-attention mechanism to fuse the information obtained from both medical sequence and graph structure data. The experimental results show that the CCDR model can effectively improve the performance of drug recommendation.

Federated learning is a new type of artificial intelligence technology. During the training process, the client transmits model parameter information instead of local data to ensure their privacy and security. But it also incurs higher communication costs. This article proposes a new federated learning pruning method, FedADP, with the aim of adaptively determining pruning ratios for each layer in each client model without infringing on client privacy, and achieving more accurate pruning effects. Our method not only reduces communication costs during the training process, but also maintains accuracy similar to the original model. We conducted experimental validation using classic models and datasets, and evaluated our scheme and traditional federated learning scheme in terms of model accuracy, communication cost, and computational cost.

Generating Adversarial Network (GAN) is a prominent unsupervised learning method that utilizes two competing neural networks to generate realistic data, which has been widely employed in image synthesis and data augmentation. Outsourcing GAN training to cloud servers can significantly reduce the computation load on local devices. Furthermore, in outsourcing settings, training data can be gathered from multiple users, leading to larger amounts of data and, as a result, improved training accuracy. However, outsourcing is associated with privacy risks, as training data often contains sensitive information. To address this problem, we propose SecureGAN, a privacy-preserving framework for GAN that aims to protect the privacy of the training input and output. We implement secure protocols based on replicated secret sharing technology to protect the privacy of the linear and nonlinear layers. We conduct experiments using the MP-SPDZ framework, and the results demonstrate the effectiveness of the proposed protocols.

Large-scale matrix multiplication is a computational bottleneck in various applications including artificial intelligence and machine learning. Given the time complexity of O(n3) for matrix multiplication, large matrix computation is exceedingly time-consuming for the client-side user. By outsourcing this task to cloud servers with substantial computational resources, we can significantly reduce the client-side computational time. This paper presents a parallel matrix multiplication outsourcing scheme based on Cannon’s algorithm. By distributing the matrix across multiple cloud servers for parallel computation, we can get a significant efficiency speedup. Our scheme employs multiple cloud servers to perform parallel matrix computation, reducing the computational load by 89-97% when utilizing 4-16 servers as opposed to using a single server. We provide a comprehensive analysis of the scheme’s correctness, security, and verifiability, substantiating the benefits of our approach through the experimental data.

Text-to-SQL aims to parse natural language problems into SQL queries, which can provide a simple interface to access large databases enabling SQL novices a quicker entry into databases. As the Text-to-SQL field is intensively studied, more and more models use GNNs to encode heterogeneous graph information in this task, and how to better obtain path information between nodes in database schema heterogeneous graphs and question-database schema heterogeneous graphs will greatly affect the effectiveness of the model parsing. Our work intends to explore the problem of solving the encoding of heterogeneous graph meta-paths in the Text-to-SQL task. Previous approaches usually use multi-layer GNNs to aggregate topological structure information between nodes. However, they ignored the structural information embedded at the edges and also failed to obtain nodes that are not directly connected but can provide contextual information through meta-paths. To solve the above problem, we propose Multi-Hop Relational Graph Attention Network based Text-to-SQL Parsing Model (MHRGATSQL) for learning topological information between nodes while obtaining semantic information embedded in the edge topology. We use multi-hop attention to modify the relational graph attention network to diffuse the attention scores throughout the network, thus increasing the “receptive field” of each layer of RGAT. Experimental results on the large-scale cross-domain Text-to-SQL dataset Spider show that our model obtains an absolute improvement of 1.7% compared to the baseline and alleviates the over-smoothing problem in the deep network model.

The proliferation of Internet of Things (IoT) devices has led to the generation of massive amounts of data that require efficient aggregation for analysis and decision-making. However, multi-tier IoT systems, which involve multiple layers of devices and gateways, face more complex security challenges in data aggregation compared to ordinary IoT systems. In this paper, we propose an efficient privacy-preserving multi-functional data aggregation scheme for multi-tier IoT architecture. The scheme supports privacy-preserving calculation of mean, variance, and anomaly proportion. The scheme uses the Paillier cryptosystem and the BLS algorithm for encryption and signature, and uses blinding techniques to keep the size of the IoT system secret. In order to make the Paillier algorithm more suitable for the IoT scenario, we also improve its efficiency of encryption and decryption. The performance evaluation shows that the scheme improves encryption efficiency by 43.7% and decryption efficiency by 45% compared to the existing scheme.

As auxiliary collaborative signals, the entity connectivity and relation semanticity beneath knowledge graph (KG) triples can alleviate the data sparsity and cold-start issues of recommendation tasks. Thus many works consider obtaining user and item representations via information aggregation on graph-structured data within Euclidean space. However, the scale-free graphs (e.g., KGs) inherently exhibit non-Euclidean geometric topologies, such as tree-like and circle-like structures. The existing recommendation models built in a single type of embedding space do not have enough capacity to embrace various geometric patterns, consequently, resulting in suboptimal performance. To address this limitation, we propose a KG-aware recommendation model with mixed-curvature manifolds interaction learning, namely CurvRec. On the one hand, it aims to preserve various global geometric structures in KG with mixed-curvature manifold spaces as the backbone. On the other hand, we integrate Ricci curvature into graph convolutional networks (GCNs) to capture local geometric structural properties when aggregating neighbor nodes. Besides, to exploit the expressive spatial features in KG, we incorporate interaction learning to ensure the geometric message passing between curved manifolds. Specifically, we adopt curvature-aware geodesic distance metrics to maximize the mutual information between Euclidean space and non-Euclidean spaces. Through extensive experiments, we demonstrate that the proposed CurvRec outperforms state-of-the-art baselines.

Face recognition is one of the key technologies in intelligent security systems. Data privacy and identification efficiency have always been concerns about face recognition. Existing privacy-preserving protocols only focus on the training phase of face recognition. Since intelligent security systems mainly complete the calculation of large-scale face data in the identification phase, existing privacy-preserving protocols cannot be well applied to intelligent security systems. In this article, we propose the first privacy-preserving face recognition protocol for the calculations in the identification phase for intelligent security systems. We introduce the Householder matrix to blind user data including model data and face data, which enables the proposed protocol to support privacy-preserving face recognition on semi-trusted edge servers. Utilizing edge computing, fast response for large-scale face recognition can be achieved. The user can offload heavy calculations of matrix multiplication and Euclidean distances to edge servers simultaneously. The proposed protocol supports parallel computing based on multiple edge servers and thus enhances the efficiency of face recognition in intelligent security systems. Moreover, the recognition accuracy in the proposed protocol is the same as that in the original PCA-based face recognition algorithm. The security analysis demonstrates that the protocol protects the privacy of user data. The numerical analysis and simulation experiments are carried out to show the efficiency and feasibility of the proposed protocol.

With the popularity of Internet of Things (IoT) and 5G, privacy-preserving message transmission and authentication have become an indispensable part in the field of data collection and analysis. There exist many protocols based on the public key cryptosystem, which allow the users to utilize their own identity as the public key to carry out data encryption and digital signature, which is very suitable for applying in the IoT environment with a large number of terminal devices. However, these protocols usually involve some complex cryptographic operations, which hinder their application on the resource-constrained IoT devices. In this paper, we design a privacy-preserving and verifiable outsourcing message transmission and authentication protocol, which allows the resource-constrained users to delegate some complex operations to the two untrusted edge servers and reduce the computational burden on the users side. The designed protocol contains several secure and novel outsourcing algorithms for modular exponentiation, bilinear pairing and scalar multiplication. For the different operations in the different situations, we design several different blinding techniques and verification methods, which not only protect the users’ private information, but also ensure the users can verify the correctness of results. Finally, we carry out some experiments to show that our proposed protocol is efficient.

With popularity and growth of cloud computing, outsourcing computation, as an important cloud service, has been applied in the field of academic and industry. It allows the resource-constrained IoT devices to outsource the computationally intensive problems to the cloud server. The smallest normalized cuts of the large-scale graph is a fundamental issue in graph theory, which is often used in various fields, such as community discovery, image segmentation, and network partition. Minimizing the normalized cuts of graph, as an NP-hard problem, can be approximately solved by the spectral decomposition. However, carrying out the spectral decomposition is very time-consuming and complicated for some IoT devices. In this article, we design a secure and efficient algorithm for outsourcing the spectral decomposition to an untrusted cloud server. We utilize a series of elementary matrices to protect both the input’s privacy and the output’s privacy from being disclosed to the cloud server. In order to ensure the correctness of the returned results, we design an efficient verification algorithm, which allows the client to detect the invalid results with a probability approximately 1. Our proposed algorithm not only reduces computational overhead on the client side, but also does not bring extra computational overhead on the cloud server side. Then, we give a theoretical analysis about correctness and privacy. Finally, we also provide some experimental results to show the feasibility of our proposed algorithm.

With the development of cloud computing and the advent of Internet of Things(IoT), outsourcing computation, as an important application of cloud computing, has been widely researched in the field of academic and industry. The convex optimization problem, as a most common mathematical problem, often appears in some machine learning algorithms and smart grid designs. However, the process of solving the convex optimization problem is very complicated and time-consuming. For some resource-constrained IoT devices, there are no enough computation resources and storage resources to deal with this problem. In this paper, we proposed an efficient and secure outsourcing algorithm for solving the large-scale convex optimization problem with equality constraints in IoT. Our proposed algorithm can not only reduce the computational complexity on the client side, but also protect the client’s sensitive data from being disclosed to the dishonest cloud server. In addition, the client can detect the malicious behavior from the cloud server with probability approximately 1. Finally, we give a theoretical analysis about correctness and security, and conduct experiments to show the feasibility of our proposed algorithm.

Data security and privacy have become an important problem while big data systems are growing dramatically fast in various application fields. Paillier additive homomorphic cryptosystem is widely used in information security fields such as big data security, communication security, cloud computing security, and artificial intelligence security. However, how to improve its computational performance is one of the most critical problems in practice. In this paper, we propose two modifications to improve the performance of the Paillier cryptosystem. Firstly, we introduce a key generation method to generate the private key with low Hamming weight, and this can be used to accelerate the decryption computation of the Paillier cryptosystem. Secondly, we propose an acceleration method based on Hensel lifting in the Paillier cryptosystem. This method can obtain a faster and improved decryption process by showing the mathematical analysis of the decryption algorithm.

Object recognition of hyperspectral remote sensing images based on machine learning is widely applied in many industries. However, the efficiency of the training and recognizing process of object recognition on hyperspectral remote sensing images is a critical issue since it involves complex matrix operations and large scale training data sets, especially for resource-constrained devices. One solution is to outsource the heavy workload of object recognition on hyperspectral remote sensing images to a cloud server. Nonetheless, it may bring some security problems when the cloud server is untrustworthy. Therefore, how to enable resource-constrained devices to securely and efficiently accomplish the training and recognizing process of object recognition on hyperspectral remote sensing images is of significant importance. In this article, we propose a secure and efficient scheme to outsource the object recognition on hyperspectral remote sensing images to the untrustworthy cloud server. The proposed scheme can protect the privacy of the computation input and output. Also, we develop an effective verification approach in our scheme that can detect the misbehavior of cloud server with the optimal probability 1. The theoretical analysis and experimental results indicate that our proposed scheme is secure and efficient.

Phrase search encryption, as an important technique in cloud-based IoT system, allows users to retrieve encrypted IoT data that contains a set of consecutive keywords. It plays an important role in cloud-based e-healthcare diagnosis system, machine learning applications for cloud-based IoT system, etc. However, to the best of our knowledge, the existing phrase search encryption schemes cannot achieve the complete verification for search results. They either cannot verify whether the returned files correctly containing the query phrase or cannot verify whether all files containing this query phrase are returned. Result verification is very important for some cloud-based IoT applications. If the search result is incorrect in the cloud-based e-healthcare diagnosis system, it will lead to misdiagnosis even endanger the patient's life. In order to deal with this problem, this article explores how to achieve verifiable phrase search over encrypted cloud-based IoT data. Specifically, we design novel look-up tables which can be utilized to determine and verify the position relationship among keywords. Meanwhile, we adopt a two-phase query strategy. In the first query phase, the data user can know the identifiers of files containing the keywords in the query phrase, and generate the search trapdoor based on these identifiers for the next phase. In the second query phase, the data user can obtain the verification information to check whether all files containing the query phrase are correctly returned. We present the security analysis of our scheme and conduct extensive experiments. The results prove the high security and efficiency of our proposed scheme.

Elliptic curve cryptosystem has been widely applied in a lot of fields, such as finance, E-commerce and E-government. In this paper, we propose an optimized FPGA implementation of elliptic curve cryptosystem over 256-bit prime fields, which has high computational performance and low resource consumption. Specifically, we design a novel modular multiplier supporting four-level pipelining, which only needs 7 clock cycles to complete a single modular multiplication. It can process 4 modular multiplication operations (4 MMPO) simultaneously. We further design, on the basis of 4MMPO, a parallel architecture to efficiently implement point doubling and point addition operation. Finally, we testify the validity of our ECC processor on Xilinx's Virtex-7 FPGA platform. The result shows that it takes only 0.15ms for an elliptic curve point multiplication, the maximum frequency of the processor is about 123.27Mhz and the resource only needs 22938 look-up tables (LUTs).

In traditional identity-based encryption schemes, security will be entirely lost once secret keys are exposed. However, with more and more use of mobile and unprotected devices, key exposure seems unavoidable. To deal with this problem, we newly propose a forward-secure identity-based public-key encryption scheme. In this primitive, the exposure of the secret key in one period doesn't affect the security of the ciphertext generated in previous periods. Any parameter in our scheme has at most log-squared complexity in terms of the total number of time periods. We also give the semantic security notions of forward-secure identity-based public-key encryption. The proposed scheme is proven semantically secure in the standard model. As far as we are concerned, it is the first forward-secure identity-based public-key encryption scheme without random oracles.

Y. D. Lyuu and M. L. Wu had proposed an improved multi-proxy multi-signature scheme, which was claimed to resist the forge attack. Lately, L. Guo and G. Wang found an inside attack on the Lyuu-Wu's scheme. In this paper, we propose a new attack on Lyuu-Wu's scheme, which can factor the parameter N and Q by using efficient number-theoretic algorithms when Q is roughly larger than the square root of N. It follows that Lyuu-Wu's scheme suffers from the forge attack from the proxy signers in that case.

How to protect the security of vital data is one of the most important issues of the database security. An efficient method is to divide the vital data into multiple parts that are stored among a group of servers by secret sharing technique. In this paper, we propose a publicly verifiable dynamic sharing protocol for data secure storage. In this protocol, the important data can be publicly verifiably shared among multiple servers, at the same time, the protocol can dynamically recover the bad shares in the system if some servers are attacked. Different from previous protocols, the new protocol is not only efficient but also practical in many circumstances because all operations can be verified by everyone not only shareholders.