UAV-to-Satellite geolocalization aims to estimate the location of an aerial-view query image taken by an Unmanned Aerial Vehicle (UAV) by matching it to satellite images annotated with known locations. However, it is difficult for existing methods to distinguish neighboring satellite images that exhibit a high degree of visual similarity. To address this issue, we introduce a plug-and-play adaptive confusion-aware contrastive loss (ACCL) to explicitly enhance the model’s discriminative ability, which gives more tolerance to high confusion query samples by means of elaborating a confusion metric function. As a plug-and-play loss module, ACCL can be easily incorporated into various UAV-to-Satellite geolocalization methods without additional modifications. To demonstrate the effectiveness of our proposed method, we conduct extensive experiments on one publicly available geolocalization dataset (i.e. NewYorkFly) and to further prove the effectiveness of our method in different scenarios, we collect two new geolocalization datasets (LasVegasFly and HollywoodFly), which contain drone-captured aerial images and dense sampled satellite images in various geomorphic regions. Experimental results indicate that our method can achieve an obvious performance improvement over the state-of-the-art methods on all three datasets. Our code and collected datasets are available at https://github.com/NWPU-CPS/ACCL.

Modular methods, which decompose complex joint policies into function-specific sub-policies, have been widely adopted to enhance asymptotic performance in single-task cooperative multi-agent reinforcement learning (MARL). However, modular policies trained on source tasks often struggle to generalize to unseen scenarios due to variations across tasks, such as mismatched action spaces and divergent state dynamics. To address this challenge, we propose Mutual Information-Guided Subtask Selection(MIGSS), a novel framework that enhances zero-shot generalization in MARL through two key innovations: a Discriminative Group Trajectory Encoder and Global Attention-Driven Coordination. Specifically, the Discriminative Group Trajectory Encoder remaps agent trajectories by maximizing mutual information between agent trajectories and dynamically assigned groups. This optimizes cross-task consistent group trajectory with broader embedding distributions. This encourages agents in distinct states to select specialized subtasks, effectively promoting functional modularity. Meanwhile, the Global Attention-Driven Coordination employs a global attention mechanism to integrate state information, coordinating group trajectories for expressive credit assignment. Extensive experiments in StarCraft II cooperative scenarios demonstrate that MIGSS significantly outperforms superior zero-shot generalization baselines in both single-task and multi-task settings.Visualization analyses confirm that the learned group trajectories successfully disperse agent trajectories into a consistent and broader embedding space, thereby enhancing subtask modularization.

Micro-expression recognition is a challenging task of identifying hidden emotion, as micro-expressions have brief durations and involve small-scale facial muscle movements. Although deep learning-based methods, especially transformer-based methods, have achieved impressive performance in this task, these methods exhibit high computational complexity and struggle to learn effective representations in the context of typically small-scale micro-expression datasets, due to the excess of tokens in the multi-head self-attention. Moreover, most existing methods do not differentiate the importance of local features, especially in micro-expression recognition with subtle changes. Therefore, we propose a novel unified Patch Tokenization, Selection and Representation framework (PTSR) with vision Transformer for micro-expression recognition. Specifically, PTSR first presents a dual norm shifted patch tokenization (DNSPT) module to learn spatial relations between neighboring pixels of the face region, which is implemented by elaborating spatial transformation and dual norm projection. Then, we employ a local-global attention module (LAM) to extract the local-global image feature, incorporating a dynamic token selection module (DTSM) to select important patches/tokens, thereby capturing more discriminative representations for the input clip. Extensive experiments are conducted on 4 widely used public datasets, i.e., CASME II, SAMM, SMIC, CAS(ME)3, and the experimental results indicate that our method can achieve clear performance improvements over the state-of-the-art methods, such as 8.37% improvement on the CAS(ME)3 dataset in terms of UF1 and 3.1% improvement on the SMIC dataset in terms of UAR metric.

Recently, diffusion models which have achieved promising progress in text-to-image generation generally have also been generally explored for image captioning. However, these diffusion-based image captioning methods usually suffer from semantic inconsistency between image content and textual description, thus producing lagging results compared with Auto-Regressive (AR) ones. To this end, in this paper, we propose a novel dual diffusion-based framework namely MirrorDiff, to achieve semantic consistency with a symmetric image-to-text-to-image generation model, which acts like a mirror that maps the original input image into a regenerated image via the generated caption. Specifically, it first utilizes both pre-trained image encoder and text encoder to obtain image representation and textual representation respectively, then forwards the image representation and the noisy textual representation into a continuous diffusion model to output an intermediate sentence. To semantically align the intermediate sentence with the input image, a diffusion-based visual regenerator is employed to regenerate the input image conditioned on the intermediate sentence, resulting in a proposed visual regeneration loss. Different from most existing image captioning methods, MirrorDiff is a plug-and-play framework which can be plugged into many previous image captioning methods, and further evaluate the generated sentence via the visual similarity between the input image and the regenerated image. Extensive experiments on the MS COCO dataset show that our method achieves obvious improvements over state-of-the-art diffusion-based methods, up to 127.9 on CIDEr, and achieves competitive performance on multiple evaluation metrics over the auto-regressive methods trained on larger-scale datasets.

Unmanned Aerial Vehicles (UAVs) are emerging as a pivotal component in the field of intelligent transportation systems. Leveraging virtual-physical interactions, digital twin technology significantly enhances the adaptability of UAVs in complex traffic environments. However, current approaches still pose three major challenges: contextual adaptability, timely responsiveness, and effective multi-UAV coordination. In this paper, we introduce EnFlexiTwin, a digital twin enhancement assistance platform seamlessly integrated with AdaSor, a lightweight adaptive data selector. EnFlexiTwin automates the construction of incremental learning datasets, enabling real-time enhancements that allow digital twins to adapt to new time series data while preserving historical knowledge. We test EnFlexiTwin on a real-world dataset from low-altitude small-parcel delivery. The results show improved performance and adaptability of digital twins. Furthermore, time-varying simulations on real-world dataset and experiments on a practical air-ground cooperative UAV swarm application highlight that EnFlexiTwin achieves superior enhancements under varying real-time requirements and swarm scale compared to baseline approaches.

The multisource universal domain adaptation (MSUDA) relaxes the constraints between the source and target domains, enabling the transfer of knowledge between domains without any restrictions on the number of source domains and the existence of unknown (private) categories. However, identifying the unknown samples in the target domain is extremely challenging since there are no available samples with the same label in source domains. Another immense challenge lies in extracting domain-invariant features for knowledge transfer since there are distribution discrepancies between each source and target domain. In this article, we propose the multirepresentation DA network (MRDAN) to classify the unlabeled targets by harnessing multiple source domains with nonidentical label sets. First, we propose a threshold-free conflict-based predictions with uncertainty (CPU) module, which comprehensively mines the complementary knowledge from different source domains to identify both known and unknown samples simultaneously. To accurately extract the domain-invariant features for recognizing known and unknown samples, a multilevel distribution alignment (MLDA) strategy is introduced to decrease the distribution discrepancy between multiple domains with nonidentical category spaces progressively. Finally, comprehensive experiments conducted on three commonly used datasets demonstrate the effectiveness of the proposed MRDAN in recognizing both known and unknown samples.

The Industrial Internet of Things (IIoT) has become a critical technology to accelerate the process of digital and intelligent transformation of industries. As the cooperative relationship between smart devices in IIoT becomes more complex, obtaining deterministic responses of IIoT periodic time-critical computing tasks becomes a crucial and nontrivial problem. However, few current works in cloud/edge/fog computing focus on this problem. This paper is a pioneer in exploring deterministic scheduling and network structural optimization problems for IIoT periodic time-critical computing tasks. We first formulate the two problems and derive theorems to help quickly identify computation and network resource sharing conflicts. Based on this, we propose a deterministic scheduling algorithm,IIoTBroker, which realizes a deterministic response for each IIoT task by optimizing the fine-grained computation and network resources, and a network optimization algorithm,IIoTDeployer, which provides a cost-effective structural upgrade solution for existing IIoT networks. Our methods are illustrated to be cost-friendly, scalable, and deterministic response guaranteed with low computation cost from our simulation results.

The shortest paths problem is a fundamental challenge in graph theory, with a broad range of potential applications. The algorithms based on matrix multiplication exhibits excellent parallelism and scalability, but is constrained by high memory consumption and algorithmic complexity. Traditional shortest paths algorithms are limited by priority queues, such as BFS and Dijkstra algorithm, making the improvement of their parallelism a focal issue. We propose a matrix operation-optimized algorithm, which offers improved parallelism, reduced time complexity, and lower memory consumption. The novel algorithm requires O(Ewcc(i)) and O(Swcc · Ewcc) times for single-source and all-pairs shortest paths problems, respectively, where Swcc and Ewcc denote the number of nodes and edges included in the largest weakly connected component in graph. To evaluate the effectiveness of the novel algorithm, we tested it using graphs from SuiteSparse Matrix Collection and Gunrock benchmark dataset. Our algorithm outperformed the BFS implementations from Gunrock and GAP (the previous state-of-the-art solution), achieving an average speedup of 3.769 × and 9.410 ×, respectively.

Vegetation parameters are important for the global carbon cycle. Therefore, the quantitative acquisition of vegetation parameters is crucial. The inverse process of the PROSAIL model has provided a classic method for vegetation parameter estimation. The current PROSAIL inverse process simulation, based on lookup tables and other methods, has challenges, such as low accuracy and poor spatial universality. To address these issues, this study proposed a PROSAIL inverse process simulation method based on deep learning. The PROSAIL model was decomposed according to the physical process of the model. Then, the corresponding network module was designed based on the inverse process of each module and combined into the IPROSAIL-Net. This network uses the reflectance of the vegetation canopies as the input and presents the leaf structure parameter, chlorophyll a + b content Cab, equivalent water thickness, dry matter content Cm, and leaf area index (LAI) as the output. This article conducted experiments using two different sets of data. PROSAIL simulation data were used to invert the five values. The inversion accuracy was above 0.99 when the number of training samples reached more than 30000. EnMAP data were used to invert the Cab and LAI values. When the number of training samples reached more than 120, the cereals accuracy was above 0.97 and the maize and rapeseed accuracies were above 0.99. The IPROSAIL-Net design was further split into six networks for separate trainings to verify its rationality. Therefore, the IPROSAIL-Net neural network is reasonable and feasible for remote sensing inversion of vegetation parameters.

Quantitative estimation of regional leaf area index (LAI) is an important basis for large-scale crop growth monitoring and yield estimation. With the development of deep learning, theoretically, the use of neural networks can effectively improve the accuracy of LAI estimation, but sufficient training samples are often required due to a large number of network parameters. In an actual regional LAI quantitative estimation, there are only a few samples, which is difficult to train in networks. Therefore, a crop dual-learning generative adversarial network (CROP-DualGAN) was proposed in this article for data enhancement of small samples to estimate regional LAI. The method uses dual learning to generate hyperspectral reflectance and corresponding LAI, including two groups of generative adversarial networks, in which the generator is used to generate data that conforms to the distribution of the training set, and the discriminator is used to judge the true or false generated samples. The generators and discriminators are constantly optimized in the confrontation so that the distribution of generated data is closer to that of training samples. In single crop type experiments, 30 training samples with enhanced in VGG16 achieved the R2of cereal, maize and rape seed as 0.921, 0.990 and 0.956, and in SSLLAI-Net achieved the R2of cereal, maize and rape seed as 0.971, 0.991 and 0.962. In multiple crop types experiments, the result is lower than individual crop estimation, but higher than that of without enhancement. Finally, non-parametric test is used to prove that most improvement in LAI estimation is significant, and the accuracy won’t decrease when improvement is not significant. In all, proposed method is universal and can effectively help benchmark models to improve regional LAI estimation accuracy with neural networks.

We analyze the prioritized sharing between an added value Mobile Virtual Network Operator (MVNO) and multiple Mobile Network Operators (MNOs). An added value MVNO is one which earns added revenue from wireless users in addition to the revenue it directly collects for providing them wireless service. To offer service, an MVNO needs to contract with one or more MNOs to utilize their networks. Agreeing on such a contract requires the MNOs to consider the impact on their revenue from allowing the MVNO to enter the market as well as the possibility that other MNOs will cooperate. To further protect their customers, the MNOs may prioritize their direct customers over those of the MVNO. We establish a multi-stage game to analyze the equilibrium decisions of the MVNO, MNOs, and users in such a setting. In particular, we characterize the condition under which the MVNO can collaborate with the MNOs. The results show that the MVNO tends to cooperate with the MNOs when the band resources are limited and the added value is significant. When there is significant difference in band resources among the MNOs, the MVNO first considers cooperating with the MNO with a smaller band. We also consider the case when the users also have access to unlicensed spectrum.

Mobile Virtual Network Operators (MVNOs) are an increasingly growing segment of the market for wireless services. MVNOs do not own their own network infrastructure and so must cooperate with existing Mobile Network Operators (MNOs) to gain access to the network infrastructure needed to enter this market. Cooperating with an MVNO is a non-trivial decision for an MNO in part because the MVNO may then become a potential competitor for customers. One motive for entering into such an arrangement is that the MVNO receives an added value from serving customers beyond what it earns from charging them for wireless service. We study a game theoretic model for the cooperation and competition between an MNO and such an added value MVNO based on models for price competition with congestible resources. Our model captures two different dimensions of how an MNO may cooperate. The first dimension is the payment scheme between the MNO and the MVNO. The second dimension is the access priority that the MNO chooses to offer to the MVNO's customers. We characterize the pros and cons of different cooperation modes and analyze the optimal cooperation mode under different conditions.

By not requiring expensive licenses, unlicensed spectrum lowers the barriers for firms to offer wireless services. However, incumbent firms may still try to erect other entry barriers. For example, recent work has highlighted how customer contracts may be used as one such barrier by penalizing customers for switching to a new entrant. However, this work did not account for another potential benefit of unlicensed spectrum, having access to this open resource may incentivize entrants to invest in new and potentially better technology. This paper studies the interaction of contracts and the incentives of firms to invest in developing new technology. We use a game theoretic model to study this and characterize the effect of contracts on economic welfare. The role of subsidies or taxes by a social planner is also considered.

A newly developed application of computed tomography (CT) is to inspect the deformations and flaws of an object in its working stage. A typical case is the CT of high-speed rotation objects (HROs), such as a working aircraft engine. Due to a relative rotation between the object and the detection equipment during a sampling time (the time for collecting each datum in the detector), directly reconstructed results by using conventional reconstruction algorithms would suffer from obvious blurring along the rotation direction. Moreover, the nonlinear relationship between measured data and a searched-for image render it difficult to remove this blur effect. Current studies on this topic mainly focus on the acquisition or the restoration of some quasi-static data. However, the mismatch between these data and a genuine image is neglected and some physical characteristics are not taken into consideration. In this paper, an innovative multi-domain regularization based general framework is proposed. Then, by using Kullback--Leibler divergence (KL-divergence) and a Poisson-Rudin-Osher-Fatemi model (Poisson-ROF or PROF) to measure the data compatibility, and total variation (TV) regularization to describe the prior knowledge of smoothness, the proposed framework is concretized into some optimization models for the HRO problem. Furthermore, correlative solution algorithms are derived and adaptive parameters (APs) are introduced to develop the method. Experiments on both simulated data and real data are provided to validate both the effectiveness and the numerical convergence of the proposed approach.

A fast ray tracing algorithm is presented. Spatial coherency is exploited by adopting a linear octree data structure which corresponds to an adaptive partitioning of space. A ray strides over a number of empty regions aligning on its way and intersects the desired objects directly, Efficiency of the algorithm is achieved by decreasing the number of regions that the ray must be checked with, by reducing the computations involved in skipping an empty region and by performing a binary search to find the next region. An efficient algorithm based on linear programming for mapping the whole environment into a sorted linear octree is also described. Only the terminal nodes containing boundary surfaces of objects are explicitly represented, which not only shortens the searching process but also leads to a considerable saving on storage space.