Person reidentification (Re-ID) is a crucial technology for intelligent security in Internet of Things (IoT) systems. Recently, unsupervised learning has been widely used for person Re-ID due to its generalization pro...
详细信息
Person reidentification (Re-ID) is a crucial technology for intelligent security in Internet of Things (IoT) systems. Recently, unsupervised learning has been widely used for person Re-ID due to its generalization property. However, the effectiveness of commonly used unsupervised clustering methods heavily relies on the quality of the clustered pseudo-labels. Moreover, pedestrian shots in real scenes are prone to factors, such as occlusion. In this article, we propose a novel global and local joint contrastive learning (GLCL) framework based on the memory bank. Specifically, we establish separate memory banks for global and local features, which are updated using global simple samples and local hard samples. The GLCL module helps excavate information from simple and hard samples, aiming to overcome the effects of poor retrieval scenarios, such as background clutter and occlusion. Additionally, we design an attributes-assisted clustering (AAC) module that utilizes pedestrian attributes to refine the clustering results. The AAC module can effectively reduce the impact of pseudo-label noise owing to the supplementary information offered by attributes. Our approach shows improved performance in person Re-ID tasks in complex scenarios, providing a promising solution for intelligent security systems in the IoT. Experimental results demonstrate the superiority of our proposed method.
Deploying task caching at edge servers has become an effectiveway to handle compute-intensive and latency-sensitive tasks on the industrialinternet. However, how to select the task scheduling location to reduce taskde...
详细信息
Deploying task caching at edge servers has become an effectiveway to handle compute-intensive and latency-sensitive tasks on the industrialinternet. However, how to select the task scheduling location to reduce taskdelay and cost while ensuring the data security and reliable communicationof edge computing remains a challenge. To solve this problem, this paperestablishes a task scheduling model with joint blockchain and task cachingin the industrial internet and designs a novel blockchain-assisted cachingmechanism to enhance system security. In this paper, the task schedulingproblem, which couples the task scheduling decision, task caching decision,and blockchain reward, is formulated as the minimum weighted cost problemunder delay constraints. This is a mixed integer nonlinear problem, which isproved to be nonconvex and NP-hard. To solve the optimal solution, thispaper proposes a task scheduling strategy algorithm based on an improvedgenetic algorithm (IGA-TSPA) by improving the genetic algorithm initializationand mutation operations to reduce the size of the initial solutionspace and enhance the optimal solution convergence speed. In addition,an Improved Least Frequently Used algorithm is proposed to improve thecontent hit rate. Simulation results show that IGA-TSPA has a faster optimalsolution-solving ability and shorter running time compared with the existingedge computing scheduling algorithms. The established task scheduling modelnot only saves 62.19% of system overhead consumption in comparison withlocal computing but also has great significance in protecting data security,reducing task processing delay, and reducing system cost.
Visual question generation task aims at generating high-quality questions about a given image. To make this tak applicable to various scenarios, e.g., the growing demand for exams, it is important to generate diverse ...
详细信息
Visual question generation task aims at generating high-quality questions about a given image. To make this tak applicable to various scenarios, e.g., the growing demand for exams, it is important to generate diverse questions. The existing methods for this task control diverse question generation based on different question types, e.g., "what" and "when." Although different question types lead to description diversity, they cannot guarantee semantic diversity when asking the same objects. Research in the field of psychology shows that humans pay attention to different objects in an image based on their preferences, which is beneficial to constructing semantically diverse questions. According to the research, we propose a multi-selector visual question generation (MS-VQG) model that aims to focus on different objects to generate diverse questions. Specifically, our MS-VQG model employs multiple selectors to imitate different humans to select different objects in a given image. Based on these different selected objects, our MS-VQG model can generate diverse questions corresponding to each selector. Extensive experiments on two datasets show that our proposed model outperforms the baselines in generating diverse questions.
In the multi-unmanned aerial vehicle (UAV) air combat confrontation environment, deriving the cooperative policy of friendly aircraft is still a challenge, owing to the higher-order differential dynamics model of airc...
详细信息
In the multi-unmanned aerial vehicle (UAV) air combat confrontation environment, deriving the cooperative policy of friendly aircraft is still a challenge, owing to the higher-order differential dynamics model of aircraft and the confidence assignment problem in multi-UAV air combat with conflict and cooperation. In this paper, a novel reinforcement learning method that combines virtual opponent and value attention decomposition is proposed. In particular, to reduce the difficulty in training induced by the higher order differential dynamics model, the actions of aircraft are abstracted into actions of the game layer and maneuvering actions of the bottom layer, in which the actions of the game layer are modeled as the pose of the virtual opponent. In the training process, only the policy of the game layer is trained, and the maneuvering policy of the bottom layer is the default policy or the rule-based policy. To address the confidence assignment problem encountered during multi-UAV cooperative training, the total value function of the team is decomposed into individual value functions based on the attention mechanism, and the policy of the game layer is optimized by integrating the individual value into the gradient computation as the baseline. Finally, the algorithm is verified on the dynamic high-fidelity training platform. The results indicate that the algorithm outperforms the state-of-the-art method in typical multi-UAV air combat scenarios such as 4V4, 5V5, and 6V6.
Large language models (LLMs) need knowledge updates to meet the ever-growing world facts and correct the hallucinated responses, facilitating the methods of lifelong model editing. Where the updated knowledge resides ...
Network topology planning is an essential multi-phase process to build and jointly optimize the multi-layer network topologies in wide-area networks (WANs). Most existing practices target single-phase/layer planning, ...
详细信息
Deep neural networks (DNNs) are vulnerable to adversarial examples with small perturbations. Adversarial defense thus has been an important means which improves the robustness of DNNs by defending against adversarial ...
详细信息
Deep neural networks (DNNs) are vulnerable to adversarial examples with small perturbations. Adversarial defense thus has been an important means which improves the robustness of DNNs by defending against adversarial examples. Existing defense methods focus on some specific types of adversarial examples and may fail to defend well in real-world applications. In practice, we may face many types of attacks where the exact type of adversarial examples in real-world applications can be even unknown. In this paper, motivated by that adversarial examples are more likely to appear near the classification boundary and are vulnerable to some transformations, we study adversarial examples from a new perspective that whether we can defend against adversarial examples by pulling them back to the original clean distribution. We empirically verify the existence of defense affine transformations that restore adversarial examples. Relying on this, we learn defense transformations to counterattack the adversarial examples by parameterizing the affine transformations and exploiting the boundary information of DNNs. Extensive experiments on both toy and real-world data sets demonstrate the effectiveness and generalization of our defense method. The code is avaliable at https://***/SCUTjinchengli/DefenseTransformer.
With the widespread use of GPS-enabled devices and services, trajectory data fuels services in a variety of fields, such as transportation and smart cities. However, trajectory data often contains errors stemming from...
详细信息
The objective of this study was to identify and synthesize functional groups for the efficient adsorption of volatile organic compounds(VOCs) through a combination of theoretical calculations,molecular design,and ex...
详细信息
The objective of this study was to identify and synthesize functional groups for the efficient adsorption of volatile organic compounds(VOCs) through a combination of theoretical calculations,molecular design,and experimental *** density functional theory(DFT) calculation,focusing on the P-containing functional groups,showed that methanol adsorption was dominated by the electrostatic interaction between the carbon surface and methanol,while toluene was mainly trapped through π-π dispersive interaction between toluene molecule and functional group *** experimental results showed the phosphorus-doped carbon materials(PCAC) prepared by directly activating potassium phytate had a phosphorus content of up to 4.5%(atom),mainly in the form of C—O—P(O)(OH)*** material exhibited a high specific area(987.6 m2·g-1) and a large adsorption capacity for methanol(440.0 mg·g-1) and toluene(350.1 mg·g-1).These properties were superior to those of the specific commercial activated carbon(CAC)sample used for comparison in this *** adsorption efficiencies per unit specific surface area of PCAC were 0.45 mg·g-1m2for methanol and 0.35 mg·g-1·m-2for *** study provided a novel theoretical and experimental framework for the molecular design of polarized elements to enhance the adsorption of polar gases,offering significant advancements over existing commercial solutions.
This paper addresses two vital challenges in Unsupervised Domain Adaptation (UDA) with a focus on harnessing the power of Vision-Language Pre-training (VLP) models. Firstly, UDA has primarily relied on ImageNet pre-tr...
详细信息
This paper addresses two vital challenges in Unsupervised Domain Adaptation (UDA) with a focus on harnessing the power of Vision-Language Pre-training (VLP) models. Firstly, UDA has primarily relied on ImageNet pre-trained models. However, the potential of VLP models in UDA remains largely unexplored. The rich representation of VLP models holds significant promise for enhancing UDA tasks. To address this, we propose a novel method called Cross-Modal Knowledge Distillation (CMKD), leveraging VLP models as teacher models to guide the learning process in the target domain, resulting in state-of-the-art performance. Secondly, current UDA paradigms involve training separate models for each task, leading to significant storage overhead and impractical model deployment as the number of transfer tasks grows. To overcome this challenge, we introduce Residual Sparse Training (RST) exploiting the benefits conferred by VLP's extensive pre-training, a technique that requires minimal adjustment (approximately 0.1%similar to 0.5%) of VLP model parameters to achieve performance comparable to fine-tuning. Combining CMKD and RST, we present a comprehensive solution that effectively leverages VLP models for UDA tasks while reducing storage overhead for model deployment. Furthermore, CMKD can serve as a baseline in conjunction with other methods like FixMatch, enhancing the performance of UDA. Our proposed method outperforms existing techniques on standard benchmarks. Our code will be available at: https://***/Wenlve-Zhou/VLP-UDA.
暂无评论