Internet of Things (IoT) applications have penetrated into all aspects human life. Millions IoT users and devices, online services, combine to create a complex heterogeneous network, which complicates the digital identity management. Distributed is promising paradigm solve problems allows soverignty over their private data. However, existing state-of-the-art methods are unsuitable for due continuing issues regarding resource limitations security privacy issues, lack systematic proof system....

Distributed backdoor attacks (DBA) have shown a higher attack success rate than centralized in federated learning (FL). However, it has not been investigated the decentralized FL. In this paper, we experimentally demonstrate that, while directly applying DBA to FL, depends on distribution of attackers network architecture. Considering that can decide their location, paper aims achieve high regardless attackers' location distribution. Specifically, first design method detect by predicting...

Blockchain has been regarded as a trusted carrier for distributed data storage. With large volumes of valuable stored on blockchain, query become major requirement. However, the existing blockchains do not provide efficient functionality because their deep-rooted chain structure. database is new direction that constructs index top blockchain to rich functionalities. The works are either insecure process separates from consensus, or inscalable all needs be in block. In this paper, we propose...

Many social studies and practical cases suggest that people's consumption behaviors are not isolated but interrelated in network services. However, most existing research either predicts users' preferences or recommends friends to users without dealing with them simultaneously. We propose a holistic approach predict on items jointly thereby make better recommendations. To this end, we design graph neural incorporates mutualistic mechanism model the mutual reinforcement relationship between...

As a decentralized trusted database, the blockchain is finding applications in growing number of fields such as finance, supply chain and medicine traceability, where large volumes valuable data are stored on blockchain. Currently, mainstream blockchains employ hybrid storage architecture combining on-chain off-chain storage. Real-time distributed search mass this system now major need. However, previous fast retrieval schemes for aimed only at without considering their relevance to data,...

Recommender systems play an essential role in providing users with accurate and positive items or services for their personalized preferences from large volume of information choices. Collaborative filtering (CF) is indispensable technique recommender widely applied many areas such as e-commerce, social medium review sites. However, CF suffers three issues which are cold start users, data sparsity. These severely degrade the recommendation performance CF. To address these issues, we propose...

In online social networks (OSNs), high trust value entities play an important role in service recommendation when users inquire certain service. Generally, OSNs are more willing to choose those services recommended by entities. fact, may suffer from great loss of property once they accept some bad provided However, current schemes do not consider this problem. Hence, we propose a scheme called RHT (recommendation entities) evaluate the degree To be specific, there exist other who provide...

With the prosperous development of Internet Things (IoT), IoT devices have been deployed in various applications, which generates large volume image data to trace and record users' behaviors, resulting better services. To accurately analyze these huge further improve experience on services, deep neural networks (DNNs) are gaining more attention become increasingly popular. However, recent studies shown that DNN models vulnerable adversarial attacks, leads risk applications practice. Previous...

Abstract The threat of Scan and Foothold Attack to the Network Edge (SFANE) is increasing, which greatly affects application development edge computing network architecture. However, existing works focus on implementation specific technologies that resist SFANE but ignore effectiveness analysis them. To overcome this limitation, paper constructs probabilistic models for evaluating edge's resistance against SFANE. In particular, attacker based ATT&CK model are first formalized. Afterward,...

The critical importance of monitoring and recognizing human emotional states in healthcare has led to a surge proposals for EEG-based multimodal emotion recognition recent years. However, practical challenges arise acquiring EEG signals daily settings due stringent data acquisition conditions, resulting the issue incomplete modalities. Existing studies have turned knowledge distillation as means mitigate this problem by transferring from networks unimodal ones. these methods are constrained...

Graph neural networks (GNNs) are a specialized type of deep learning models on graphs by aggregations over neighbor nodes. However, recent studies reveal that the performance GNNs severely deteriorated injecting adversarial examples. Hence, improving robustness is significant importance. Prior works devoted to reducing influence direct adversaries which attacks positioning node's one-hop neighbors, yet these approaches limited in protecting from indirect within multi-hop neighbors. In this...

Deep learning-based image recognition technology has significantly advanced the development of modern industrial intelligence. However, issue adversarial examples that follows gradually garnered attention researchers. By injecting disturbances are difficult for humans to detect into image, deep learning model generates incorrect results, severely impacting security applications. To address this problem and improve accuracy robustness models, a combined detection method based on residual...

The ability to decentralize knowledge graphs (KG) is important exploit the full potential of Semantic Web and realize 3.0 vision. However, decentralization also renders KGs more prone attacks with adverse effects on data integrity query verifiability. While existing studies focus ensuring integrity, how ensure verifiability - thus guarding against incorrect, incomplete, or outdated results remains unsolved. We propose VeriDKG, first SPARQL engine for decentralized (DKG) that offers both...

Session-based recommendation is a challenging field in the research network-based behavior modeling, mainly due to complex transfer of user interests between items and limited information. The previous methods model session as sequence or graph, which takes into account role time attention session-based achieve satisfactory performances, they still ignore latent relationship interest item category. In this study, we propose category attentive graph neural network (CAGNN) for recommendation....