Text-to-image (T2I) diffusion models have become popular in computer vision, but they remain vulnerable to backdoor attacks. Existing methods typically trigger a fixed image regardless of user input, causing severe semantic inconsistency between the generated image and the original prompt. This makes the attack easily detectable by machines as it would lack visual stealth. To overcome this challenge, we propose MultiAttack, a novel semantic-preserving multi-object coexistence backdoor attack for T2I diffusion models, which retains prompt-described objects while injecting malicious targets. First, we propose a semantic-preserving data poisoning strategy to build a latent mapping, which maps the trigger into a composite semantic space while retaining the original prompt context. Second, we design a backdoor enhancement mechanism to embed the spatial orthogonality between malicious and benign objects into model weights as a conditional response, which strengthens the model’s ability to generate stable malicious outputs without requiring additional inference. Results on Stable Diffusion show that compared tostate-of-the-art baselines, MultiAttack increases attack success rate by 13.1% and visual stealth (defined as the success rate of co-generating both prompt-described and target objects) by 12.6%, with an FID increase of less than 1.2 and a CLIP score decrease of under 1 compared to clean models.

In recent years, with the rapid popularization of smart consumer electronics, privacy and security issues faced by consumers when accessing smart electronic devices have increasingly drawn attention. In resource-constrained consumer electronic devices, explicit access policies can trigger significant privacy risks. To address this problem, researchers have proposed a novel ciphertext-policy attribute-based encryption (CP-ABE) scheme known as the hidden policy CP-ABE scheme (HP-CP-ABE), which hides the access structure while maintaining data confidentiality. However, the existing HP-CP-ABE schemes face two major challenges: unlimited access privileges may lead to malicious users obtaining sensitive data and compromising the privacy of other users, and the high computational overhead is difficult to adapt to constrained consumer electronic devices. This paper proposes a time-limited and server-aided CP-ABE (TS-ABE) scheme with a hidden policy. This scheme achieves constraints on access control through time-limited access and significantly limits data exposure risks within strictly defined time periods, thereby protecting consumer privacy; meanwhile, computationally intensive tasks are offloaded to a multi-server assistance system framework, enabling lightweight operation on resource-constrained consumer electronic devices. Performance analysis and evaluation demonstrate that the TS-ABE scheme offers notable computational efficiency and storage overhead benefits.

Graph neural networks (GNNs) have been widely applied to various machine learning tasks, particularly in social network analysis and healthcare. Meanwhile, privacy concerns in GNN have attracted increasing attention. In graph-level classification tasks, differential privacy mechanisms are commonly employed to protect sensitive information, often implemented through stochastic gradient descent. However, the addition of excessive noise can significantly degrade model performance. In this paper, we propose layer-adaptive differential privacy (LADP), an adaptive noise injection mechanism designed for graph classification tasks. In the proposed mechanism, the noise addition process is separated into the stages of node feature aggregation and whole-graph feature update. To reduce the impact of noise on model utility, adaptive noise is added to each layer individually, allowing for more precise gradient control across the network. We evaluate our method on several public benchmark datasets. Experimental results show that LADP achieves higher classification accuracy than existing approaches under comparable privacy budgets.

Crowded object detection is a highly challenging task in the field of object detection. One major issue is that the standard NMS algorithm encounters difficulties in crowded scenes, as it often erroneously suppresses highly overlapped predictions. In fact, if there are highly clustered bounding boxes in a region, they usually correspond to a ground-truth object. Based on this pattern, we propose an effective cluster-NMS algorithm which can enhance the confidence of not being suppressed for a highly scored bounding box by taking into account the surrounding clustered boxes. It ensures that true predictions are correctly retained in the non-maximum suppression step. On the CrowdHuman dataset, cluster-NMS outperforms other NMS algorithms in improving the performance of crowded object detection. The YOLOv7 detector with cluster-NMS achieves 91.4

Object detection in drone-captured scenarios presents significant challenges due to factors such as varying object scales, motion blur, and dense object clusters. Although existing methods, including attention blocks and feature fusion networks, have shown improvements in detection accuracy, they often come with high computational costs, which hinder realtime performance. In this paper, we propose IFN-YOLOv8, an enhanced version of YOLOv8, designed to address these challenges. By integrating the P2 feature scale, IFN-YOLOv8 enhances small object detection through higher-resolution feature maps. Additionally, we introduce a novel convolutional block, RHAConv, to replace traditional convolution layers, improving feature representation in scenes with dense object clusters. A new Information Fusion Module is also proposed to refine object features, reducing both missed and false detections. Experimental results on the VisDrone and DOTA datasets demonstrate that IFN-YOLOv8 outperforms mainstream methods, achieving an mAP@50 of 45.7% and 68.5%, respectively, while maintaining low resource consumption and high detection speed.

Object detection in crowded scenes involves various difficulties, such as small objects, occluded objects, and insufficient features. Existing models for crowded object detection often focus on only one detection difficulty, and they are too large to be applied in practice. To address the diverse challenges of object detection in crowded scenes, we construct a lightweight crowded object detector called MBB-YOLO, which contains several modules for comprehensive improvement. To improve the network's ability to extract fine-grained features, we use SPD-Conv and the proposed MS-Conv to replace the strided convolution in the network. An bi-branch multi-scale convolution attention (BMCA) module is proposed to aggregate multi-scale contextual information. We also propose boundary-NMS to better identify proposal boxes from different objects, which reduces suppression errors caused by object occlusion. MBB-YOLO achieves 87.6% AP and an inference speed of 78.8 FPS on the CrowdHuman dataset, which surpasses other mainstream lightweight object detectors.

With the proliferation of the Internet of Things (IoT), access control has become particularly important in protecting data security. However, traditional access control models have limited capabilities in addressing challenges such as single points of failure, coarse-grained access control, and insufficient security in distributed environments. To tackle these challenges, we propose DTSAC, a smart contract-based access control with distributed delegation and dynamic trust management. This model utilizes blockchain for the decentralized delegation of capabilities and distributes delegation relationships across multiple delegation trees to enable fast revocation of capabilities. In addition, we design a dynamic trust assessment mechanism between the delegator and the delegatee to more comprehensively evaluate the credibility of the subject. Finally, a functional prototype of the DTSAC is successfully deployed on the public Sepolia Ethereum testnet. The experimental evaluation is conducted to verify the feasibility and effectiveness of the proposed model.

Small object detection is one of the important tasks in object detection. Compared with the whole image, the small object has a low proportion, and the feature information of the object is very little, which brings a huge challenge to small object detection. In order to address the issue, we propose a small object detection algorithm based on split mixed attention, named as SMAMYOLO. Firstly, based on the channel attention module and the spatial attention module, we recombine their connection structures and propose a mixed attention module, named as MAM which can enhance the model’s ability to express small object features in spatial dimension. Secondly, according to the effects of different receptive field sizes on the object, we propose a split feature extraction module based on mixed attention, named as SMAM which can adaptively adjust the size of the receptive field according to the scale of the input feature map, and use mixed attention to enhance the ability of different branches to capture small object feature information.Finally, based on SMAM and CSPNet, the core residual module in yolov5 is improved, and a feature extraction module, named as CSMAM, is proposed. The additional computational overhead of CSMAM can be ignored. Based on the TinyPerson dataset, the experiments have been conducted. Firstly, the ablation experiment shows that the average detection accuracy of SMAM-YOLO algorithm is improved by 4.15% compared with the baseline algorithm, and the FPS reaches 74 frame/s; In addition, compared with some existing mainstream small target detection models, the experimental results show that the SMAM-YOLO algorithm improves the detection accuracy by 1.46%～6.84% on average, and can meet the requirements of real-time detection.

在新时代,课程思政是一种新的教育理念.《Python语言程序设计》是高校广泛开设的程序设计类公共通识课程,其课程思政的重要性不言而喻.首先分析了课程思政的必要性;然后,以财务管理专业《Python语言程序设计》课程教学为例,探讨了《Python语言程序设计》课程思政的教学目标,深度挖掘知识点中蕴含的思政元素,精心设计课程思政典型案例,在知识传授过程中自然融合思政元素,起到润物无声的育人效果;最后,对课程思政的效果进行评价和分析.

In order to protect private data in edge computing system and ensure that ciphertexts encrypted under different keys can be processed in the cloud, we propose a privacy protection protocol for edge computing, which is based on blockchain technology and multikey fully homomorphic encryption (MFHE). By constructing a Merkle hash tree and saving the Merkle root node to the blockchain, the integrity of the data in edge nodes is protected. For the data with high computational complexity in edge nodes, the GSW fully homomorphic encryption scheme is employed to process the encrypted private data of a single edge node by the CMS. The LinkAlgo algorithm is applied to link different keys by the relation between them, which enables the CMS to process the encrypted private data of multiple edge nodes. The security analysis and performance evaluation of the proposed privacy protection protocol show that our protocol has high security and good efficiency.

Since detection and recognition of traffic signs is one of three modules of autonomous driving, it has been concerned by many researchers. But the traffic sign images, which are gathered in complex environments such as bright or foggy weather, are often obscured or blurred. This may bring about serious challenge to recognize traffic sign. In this paper, a novel and adaptive traffic sign recognition scheme is proposed, which addresses the issue of traffic sign recognition in complex environments. First, according to the environmental conditions, the traffic sign images are classified into four categories by Swin Transformer. Then, based on INDANE and ACE algorithms, an adaptive dataset enhancement algorithm is proposed to enhance the classified dataset and strengthen the image features. Finally, yolov5 algorithm is improved to effectively accomplish traffic sign recognition by reducing the maximum down sampling multiple, adjusting the stacking proportion of residual blocks, increasing the convolution kernel, and replacing the normalization method. Based on the GTSRB dataset, the experiments are conducted to evaluate the performance of the proposed scheme. The experimental results show that the mAP of our proposed scheme is 99.86% when traffic sign is recognized in complex environment. Compared with the existing schemes, our proposed scheme has higher recognition accuracy and wider application.

Differential privacy is an effective measure of privacy protection in data analysis. We propose a differential private trajectory data publication method based on consistency constraints in road network space to significantly improve the accuracy of a general class of trajectory statistics queries. First of all, Laplace noise is injected into statistical data of each road segment. And then, in the post-processing phase, consistency constraint is employed to hold over the noisy output. Based on both synthetic datasets, we do experiments to evaluate the performance of the proposed method. The experimental results show that the proposed method achieves high availability and efficiency.

Due to the transparency of the blockchain, the data in the blockchain can be viewed by any joining node, and the privacy is weak. In order to better solve the problem of privacy protection in the current blockchain, for the application of e-commerce blockchain, the Pedersen commitment mechanism is adopted to hide the transaction amount, combined with zero-knowledge proof technology to realize the verification of the transaction amount. At the same time, an efficient range proof scheme based on polynomial commitment and vector inner-product commitment is used to verify whether the transaction amount and the balance of both parties are greater than zero. Finally, use the homomorphism promised by Pedersen to update the ciphertext ledger. The security analysis and efficiency test of the proposed blockchain privacy protection algorithm show that compared with the existing schemes, the proposed algorithm has the advantages of strong security and higher efficiency.

At present the rapidly growing malicious applications in Android systems have imposed significant threats to the security of mobile terminals,but the traditional detection systems fail to detect them quickly and effectively.To address the problem,this paper proposes a malware detection system which combines feature weighting with the deep learning algorithm using Bidirectional Long Short-Term Memory (Bi-LSTM) neural network.The static analysis method is used to extract different types of behavior features from malicious and normal applications.The feature weighting method is used to eliminate noise and irrelevant factors to construct feature vectors.The Bi-LSTM-based deep learning algorithm is used to optimize the behavior feature parameters.Then a classification model for malicious and normal applications is designed,and on this basis a detection system for malicious applications combining feature weighting and the deep learning algorithm is constructed.Experimental results show that compared with traditional detection systems such as Support Vector Machine(SVM) and RNN,the proposed system has higher precision and accuracy in malicious application detection.

With the development of technology and the increasing popularity of smart devices, more and more people use social network sites. When users express their opinions on the Internet, their personal privacy may be inadvertently exposed, which make the privacy issue more obvious. At present, many existing studies consider methods to encrypt or calculate privacy ratings without considering whether this is unreasonable for social networks. Meanwhile, these studies ignore the limitation of memory resources. In order to protect the privacy of sensitive information with limited resources, classification of privacy information is particularly important. In this paper, we discuss a privacy security classification model for the attributes of online social network users. Through this method, the privacy risk degree of user attributes on social network can be clearly understood. In addition, we introduce visibility innovatively into the privacy security classification model and provide a reference for future research.

With the increasing number of vehicles, traffic management has put forward higher requirements for vehicle monitoring, thus the technology of vehicle detection based on drone images has received increasing attention. Firstly, we construct a new vehicle detection data set of 600 drone images so that to solve the vehicle detection tasks in real world. Secondly, aiming at the problem of false detection and missed detection in vehicle detection, the Faster R-CNN is improved by using ResNet and constructing Feature Pyramid Networks (FPN) to extract the image features. Finally, based on the vehicle detection data set, the improved Faster R-CNN can be used to detect vehicle targets. The experiment results show that the accuracy of improved method is 96.83%, which is 3.86% higher than that of the original Faster R-CNN method.

In today’s society, Global Mobile Networks (GLOMONETs) have become an important network infrastructure that provides seamless roaming service for mobile users when they leave their home network. Authentication is an essential mechanism for secure communication among the mobile user, home network, and foreign network in GLOMONET. Recently, Madhusudhan and Shashidhara presented a lightweight authentication protocol for roaming application in GLOMONET. However, we found their protocol not only has design flaws, but is also vulnerable to many attacks. To address these weaknesses, this paper proposes a novel authentication protocol with strong security for GLOMONET based on previous work. The fuzzy verifier technique makes the protocol free from smart card breach attack, while achieving the feature of local password change. Moreover, the computational intractability of the Discrete Logarithm Problem (DLP) guarantees the security of the session key. The security of the protocol is verified by the ProVerif tool. Compared with other related protocols, our protocol achieves a higher level of security at the expense of small increases in computational cost and communication cost. Therefore, it is more suitable for securing the roaming application in GLOMONET.

The Internet of Things (IoT) is an emerging technology and expected to provide solutions for various industrial fields. As a basic technology of the IoT, wireless sensor networks (WSNs) can be used to collect the required environment parameters for specific applications. Due to the resource limitation of sensor node and the open nature of wireless channel, security has become an enormous challenge in WSN. Authentication as a basic security service can be used to guarantee the legality of data access in WSN. Recently, Chang and Le proposed two authentication protocols for WSN for different security requirements. However, their protocol cannot provide proper mutual authentication and has other security and functionality defects. We present a three-factor user authentication protocol for WSN to remove the weaknesses of previous protocols. The security of the proposed protocol is analyzed, and the security, functionality and performance of our protocol are compared with other related protocols. The comparison results and simulation results by NS-3 show that the proposed protocol is robust and energy efficient for IoT applications.

With the development of wireless communication and positioning technology, location-based services (LBSs) have been gaining tremendous popularity, due to its ability to greatly facilitate the people's daily lives. Meanwhile, it also entails the risk of location privacy disclosure. To address this issue, we usually adopt K-anonymity in the centralized architecture based on a single trusted anonymizer. However, it may expose the user's privacy in continuous LBSs. In this paper, we propose a dual-K mechanism (DKM) to protect the user's trajectory privacy for continuous LBSs. Our scheme introduces multiple anonymizers between the user and the location service provider (LSP), and each time the K query locations are sent to K anonymizers to form K-anonymity respectively. At the same time, we combine with location selection mechanism to confuse the user's real location to enhance the user's privacy. The LSP and a single anonymizer cannot get the user's trajectory, and the anonymizers can be semi-trusted. The security analysis demonstrates that our scheme can effectively protect the user's trajectory privacy.

近年来,随着定位技术和无线移动领域相关技术的飞速发展,基于位置的服务得到广泛应用.但是,在使用基于位置的服务时,可能使位置信息暴露,甚至导致个人隐私的泄露.提出一种基于空间区域匿名的位置隐私保护方案—LocPriv,方案采用中心服务器模型,使用假名匿名技术和假位置匿名技术实现位置隐私保护,对方案的匿名性和性能进行分析,并通过实验进行验证.LocPriv具有较高的匿名性,能够保护用户的位置隐私,同时降低了整个系统的通信开销和响应时间,在一定程度上避免了区域中心遭受攻击.