Unmanned Aerial Vehicle (UAV) video surveillance has become indispensable for intelligent transportation systems, autonomous driving and unmanned vehicle applications. However, traffic objects captured from high altitudes typically exhibit small sizes and weak feature representation, while being susceptible to interference from UAV motion and complex road environments, resulting in detection challenges. Additionally, multi-scale object detection requires a balance between accuracy and efficiency. To address these issues, we propose a real-time small traffic object detection method based on adaptive fine-grained attention and multi-scale fusion mechanism. Building upon the Real-Time Detection Transformer (RT-DETR), we construct a Small Object Enhanced Real-Time Detection Transformer (SOE-RTDETR) with an adaptive fine-grained channel attention block to enhance small object feature extraction. The SOE-RTDETR incorporates CSP-Dilated Reparam Residual Blocks (CSP-DRRB) that expand the convolutional receptive field while maintaining computational efficiency through depthwise convolution and reparameterization. We further optimize the attention-based intra-scale feature interaction module using deformable attention mechanism and propose a bidirectional feature pyramid network specifically designed for small traffic object detection to strengthen multi-scale feature fusion. Experimental results on the VisDrone2019 dataset demonstrate that the proposed SOE-RTDETR achieves improvements of 2.9% in mAP@50 and 2.3% in mAP@50:95 on the validation set, and 2.3% and 1.5% on the test set compared to the baseline RT-DETR-r18 model, while maintaining equivalent model size and computational cost.

Gaussian processes (GPs) have attracted considerable attention in assisting evolutionary algorithms (EAs) to solve computationally expensive optimization problems (EOPs) because they can directly provide information about the uncertainty of their predictions. However, the computational complexity of GPs grows cubically as the amount of data increases, which severely limits their computational efficiency in high-dimensional expensive multiobjective optimization problems (EMOPs). To address this limitation, we propose a surrogate-assisted evolutionary algorithm (SAEA) that integrates multilayer perceptron (MLP) grouping with sparse GPs, referred to as MLPSGP-SAEA. First, the MLP grouping selects a subspace from the original space by evaluating the impact of each decision variable on the objective functions. Then, for each objective function, a sparse GP model is employed, and the locations of pseudo-input points are optimized to enhance computational efficiency while improving model accuracy. Moreover, an adaptive sparse and diverse (ASD) infill criterion is proposed, based on the characteristics of the sparse GP model predictive distribution, to better balance exploration and exploitation. Finally, extensive experiments are conducted on four benchmark suites and an aerodynamic design optimization problem. The experimental results demonstrate that MLPSGP-SAEA exhibits significant competitive advantages over the state-of-the-art SAEAs.

ABSTRACT The Density Peaks Clustering (DPC) algorithm is well‐known for its simplicity and efficiency in clustering data of arbitrary shapes. However, it faces challenges such as inconsistent local density definitions and sample assignment errors. This paper introduces the Shared Neighbors and Natural Neighbors Density Peaks Clustering (SN‐DPC) algorithm to address these issues. SN‐DPC redefines local density by incorporating weighted shared neighbors, which enhances the density contribution from distant samples and provides a better representation of the data distribution. It also establishes a new similarity measure between samples using shared and natural neighbors, which increases intra‐cluster similarity and reduces assignment errors, thereby improving clustering performance. Compared with DPC‐CE, IDPC‐FA, DPCSA, FNDPC, and traditional DPC, SN‐DPC demonstrated superior effectiveness on both synthetic and real datasets. When applied to the analysis of electricity consumption patterns, it more accurately identified load consumption patterns and usage habits.

ABSTRACT Aiming to address issues such as non‐uniform rain density and misjudgment caused by noise in image de‐raining, we propose a single‐image de‐raining method based on a generative adversarial network with contextual information aggregation and multi‐scale feature fusion. First, we design a generator composed of encoding, context information aggregation, and decoding stages. Features are extracted using convolution, while expansion convolution effectively aggregates context information. Transposition convolution is then used to restore the image, enhancing the model's ability to perceive image details and achieve accurate image information judgment and content reconstruction. Second, we design a multi‐scale feature fusion discriminator structure to capture different image details using convolution kernels of different scales and connect feature maps from different scales. This improves the model's ability to understand image details and differentiate between authentic and fake images. Finally, we propose a new refinement loss function to reduce grid artifact generation and add Lipschitz constraints to further minimize the imaging gap. In this paper, peak signal‐to‐noise ratio and structural similarity are used as evaluation criteria, and experiments conducted on real and synthesized rain maps demonstrate the superior rain removal performance of the proposed method.

The combination of Cloud Radio Access Network (C-RAN) and Mobile Edge Computing (MEC) has been proven to effectively enhance network transmission rates and service capabilities. However, since the ground communication facilities in the architecture are fixed, have limited coverage and are easily damaged by natural disasters, some user requests may not be processed in a timely manner. This poses a significant challenge to the flexibility and post-disaster recovery capabilities of the architecture. To this end, this paper studies how to deploy Unmanned Aerial Vehicles (UAVs) to assist the recovery of post-disaster communication network when some ground facilities in the architecture are destroyed, and how to perform reasonable task scheduling based on user task priorities to achieve rapid and effective rescue. Moreover, this paper mathematically models the problem with the goal of maximizing the success rate of user task execution. Since the established model belongs to the Mixed-integer Nonlinear Programming (MINLP) model, which is non-convex and NP-Hard, this paper designs a priority-based algorithm for joint UAV deployment and task scheduling to obtain high-quality suboptimal solutions. The algorithm employs a two-layer optimization architecture with the characteristics of low memory usage and low time complexity. The experimental results indicate that the proposed algorithm outperforms the baseline algorithms and is more effective in handling rescue tasks under different user scales.

Summary Accurate short‐term load forecasting plays an important role in reducing power generation costs, maintaining supply and demand balance, and stabling the power grids operation. In recent years, deep learning models based on recurrent neural networks (RNN) have been widely used in short‐term load forecasting. Nevertheless, RNN cannot extract multi‐scale features of load data, resulting in low forecasting accuracy. A model for short‐term power load forecasting of residual multiscale‐RNN (RM‐RNN) was proposed in this study. RM‐RNN uses the multilayer RNN network structure. Specifically, each layer sets the dilated convolution with different dilated coefficients to extract the multi‐scale features of the load data. Adjacent networks transfer feature information for feature fusion through the residual structures. The experiment used random sampling data training model, and compared RM‐RNN with multiple deep learning models. The experimental results demonstrated that the mean error of RM‐RNN prediction is the lowest, indicating that dilated convolution can effectively extract multi‐scale features of load data. This result verified the effectiveness of residual structure fusion features, and improved the accuracy of short‐term load forecasting.

Hyperspectral image (HSI) has received considerable attention in the field of target detection due to its powerful ability to capture the spectral information of land covers, and plenty of detection algorithms have been explored. However, these methods generally leverage the difference between the spectrum of the target to be detected and the background spectrum to accomplish target detection, and so are susceptible to the problem of spectral variability. In this article, we propose a global-to-local hierarchical detection algorithm for HSI (G2LHTD). Firstly, extended morphological attribute profile (EMAP) is first used to model global spatial texture information from HSI. Subsequently, a diverse-direction constrained energy minimization (D2CEM) detector is developed to consider the spatial information within eight neighborhoods around each pixel in HSI, yielding comprehensive local spatial information. More substantially, to effectively discriminate the neighborhood information in diverse directions, we devise an adaptive neighborhood feature aggregation (ANFA) strategy, which will comprehensively evaluate the significance of neighborhood information in diverse directions. As a result, the spatial features of HSI can be comprehensively considered for hyperspectral target detection (HTD). Extensive experiments, conducted on four standard datasets, demonstrate the effectiveness of the proposed method. The codes of this work will be available at https://github.com/zhonghaocheng/G2LHTD_Master for the sake of reproducibility.

A single strategy used in the firefly algorithm (FA) cannot effectively solve the complex optimal scheduling problem. Thus, we propose the FA with division of roles (DRFA). Herein, fireflies are divided into leaders, developers, and followers, while a learning strategy is assigned to each role: the leader chooses the greedy Cauchy mutation; the developer chooses two leaders randomly and uses the elite neighborhood search strategy for local development; the follower randomly selects two excellent particles for global exploration. To improve the efficiency of the fixed step size used in FA, a stepped variable step size strategy is proposed to meet different requirements of the algorithm for the step size at different stages. Role division can balance the development and exploration ability of the algorithm. The use of multiple strategies can greatly improve the versatility of the algorithm for complex optimization problems. The optimal performance of the proposed algorithm has been verified by three sets of test functions and a simulation of optimal scheduling of cascade reservoirs.

The rapid growth of video traffic poses serious challenges to the current Internet. Content‐Centric Networking (CCN) as a promising candidate has been proposed to reengineer the Internet architecture. The in‐network caching and named content communication model of CCN can enhance the video streaming applications and reduce the network workload. Due to the bandwidth‐consuming characteristic of video streaming, the aggressive transmission of video data will cause a reduction of overall network efficiency. In this paper, we present an adaptive video transmission mechanism over Mobile Edge Computing‐ (MEC‐) based CCN. The computation and storage resources of the MEC server are utilized to facilitate the video delivery. Our mechanism adopts a scalable video coding scheme to adaptively control transmission rate to cope with the network condition variation. To analyse the equilibrium property of the proposed mechanism, an analytical model is deduced by using network utility function and convex programming. We also take into account the packet loss in wired and wireless links and present a MEC assistant loss recovery algorithm. The experiment results demonstrate the performance improvement of our proposed mechanism.

Summary In density peaks clustering, its complexity for computing local density and relative distance of samples raises a scalability issue for processing large datasets. To address the issue, density peaks clustering based on circular partition and grid similarity has been proposed. The algorithm partitions the data space into circular grids, with each grid treated as a sample, for determining the number of clusters and searching for the density peaks; then, a new grid similarity is calculated to effectively assign unallocated grids. The proposed circular partition method effectively reduces the number of samples and the computational complexity. Extensive experiments have been conducted on several datasets with arbitrary shapes and scales, and the proposed method outperforms other density peaks clustering variants in terms of clustering accuracy and efficiency.

Firefly algorithm (FA) is a population-based stochastic algorithm, which is inspired by the behavior of the flashing of fireflies. Though some recent studies show that FA is effective on many optimization problems, its performance is greatly influenced by its control parameters. In this paper, a new FA called adaptive FA with alternative search (AFAas) is proposed to improve the performance of FA. The main contribution of this paper consists of two aspects: 1) an adaptive strategy is used to dynamically adjust the control parameters; and 2) an alternative search strategy is employed to enhance the global and local search abilities. Experiments are conducted on a set of well-known benchmark functions. Computational results show that our approach AFAas achieves better solutions than the standard FA and some recently published FA variants.