Automatic underwater vehicle (AUV) object detection in sonar images has drawn widespread interest for its practical applications. Nevertheless, accurate detection remains challenging due to the scale imbalance and arbitrary orientations of underwater objects with extreme aspect ratios. In this article, we introduce a refined multi-scale feature enhancement network (RMFENet) framework as a novel approach to achieve high-precision performance in detecting arbitrary-oriented underwater objects. Firstly, we develop a data preprocessing module to adaptively adjust large-size sonar images to the appropriate size and direction angle. Secondly, we construct a composite backbone network (CB-Net) that refines the integration of high-level and low-level features in sonar images and enhances the correlation of different features. Thirdly, the shuffle convolutional block attention module (Shuffle-CBAM) mechanism and diverse multi-scale feature fusion block (DMSFF) further enhance the model feature extraction ability. Finally, to generate more accurate rotation region proposals and anchors, we propose a rotated intersection over union (RIoU) to solve inaccurate object region localization problems and high computational complexity in existing methods. Besides, our model is constructed on two sonar image datasets for oriented underwater object detection with different spatial resolutions, diverse object categories, and bounding box annotation. Extensive experiments demonstrate that our method can achieve state-of-the-art performance on the constructed sonar image object detection datasets while keeping better robustness.

Identifying the association between long noncoding RNA (lncRNA) and micro‐RNA (miRNA) is of great significance for the treatment of diseases by interfering with the combination of miRNA and messenger RNA (mRNA). Although many efforts and resources have been invested to identify lncRNA‐miRNA associations (LMAs), clinical trials are still expensive and laborious. Nevertheless, the experiments also need to consult a large number of side effects. Therefore, novel computer‐aided models are urgently needed to predict LMAs. This paper proposed a semantic embedded bipartite graph network for predicting lncRNA‐miRNA associations (SEBGLMA), which provided a novel feature extraction method by integrating K‐mer segmentation, word2vec, Gaussian interaction profile (GIP), and graph convolution network (GCN). Concretely, the attribute characteristics of RNA sequences are extracted by K‐mer segmentation and word2vec modules. Afterward, the adjacent matrix is completed by GIP self‐similarity. Then, the attribute characteristics and adjacent matrix are fed into GCN for embedding behavior features. Finally, the features are sent into the rotation forest (RoF) for detecting potential LMAs. The average accuracy, precision, sensitivity, specificity, Matthews correlation coefficient, and F1‐Score are 87.09%, 87.66%, 87.03%, 87.84%, 74.18%, and 86.99% on the benchmark data set. For fairly validating the performance of our model, we conducted various comparisons with different classifiers. Furthermore, the case studies of hsa‐miR‐497‐5P and NONHSAT022145.2 are also established. The results of comparisons and case studies further illustrated that our method is anticipated to become a robust and reliable tool for the identification of LMAs.

Compared with side-scan sonar (SSS) and forward-looking sonar (FLS), synthetic aperture sonar (SAS) device can generate high-resolution underwater images, which is important for marine topographic mapping. Nevertheless, existing deep learning methods face challenges in extracting detailed feature information from underwater SAS images for semantic segmentation tasks, primarily due to the significant interference from the complex underwater environment and seabed reverberation noise. To address these challenges, we propose a novel network named U-SAS for SAS image semantic segmentation, which uses the hybrid convolution and attention mechanism architecture to extract rich multi-level features from SAS images. Specifically, U-SAS separates the extracted features into complementary representation, namely salient features and abstract location features. To enhance the distinctive representation of multi-level features, U-SAS incorporates the max-min module (MMM) and convolutional block attention mechanism (CBAM). The MMM effectively emphasizes the robust and significant underwater acoustic features, while the CBAM suppresses seabed reverberation noise and enhances the representation of positional information. Besides, we construct the global decoding module (GDM) to fuse salient features and abstract location features to enhance the correlation between local features and achieve global semantic understanding. To verify the effectiveness and feasibility of the proposed U-SAS network, we conducted extensive experiments on the SAS image dataset of complex underwater scenes. The experimental results show that U-SAS achieves 78.58% and 58.59% of mAcc and mIoU, respectively, which outperforms other state-of-the-art methods. Implementation codes will be available on https://github.com/darkseid-arch/U-SAS.

With the rapid growth of global air traffic, 4D-flight trajectory prediction (4D-FTP) using deep learning (DL) methods has become essential for applications such as flight delay prediction, fuel consumption analysis, and traffic management. However, the adversarial attacks pose significant security threats to DL-based 4D-FTP systems reliant on automatic dependent surveillance-broadcast (ADS-B) sensors. Furthermore, existing vulnerabilities in time series prediction (TSP) models for 4D-FTP remain underexplored due to the lack of effective and stealthy attack methods. To address this gap, we propose the Dynamic Quasi-Hyperbolic Momentum Iterative Attack with Small Perturbation (DQM-Attack). This method leverages quasi-hyperbolic momentum (QHM) and dynamic step sizes to optimize gradient utilization in 4D-FTP models. In addition, attack stealth is enhanced through the Manifold Smooth Module (MSM) and Sparse Smooth Reinforcement Learning (SS-RL). Experimental results demonstrate that DQM-Attack effectively disrupts predictions with minimal perturbations across four state-of-the-art TSP models. This study appears to be the first to address stealthy adversarial attacks on 4D-FTP systems, revealing a critical security vulnerability in air traffic management (ATM).

Identifying the interactions of the drug‐target is central to the cognate areas including drug discovery and drug reposition. Although the high‐throughput biotechnologies have made tremendous progress, the indispensable clinical trials remain to be expensive, laborious, and intricate. Therefore, a convenient and reliable computer‐aided method has become the focus on inferring drug‐target interactions (DTIs). In this research, we propose a novel computational model integrating a pyramid histogram of oriented gradients (PHOG), Position‐Specific Scoring Matrix (PSSM), and rotation forest (RF) classifier for identifying DTIs. Specifically, protein primary sequences are first converted into PSSMs to describe the potential biological evolution information. After that, PHOG is employed to mine the highly representative features of PSSM from multiple pyramid levels, and the complete describers of drug‐target pairs are generated by combining the molecular substructure fingerprints and PHOG features. Finally, we feed the complete describers into the RF classifier for effective prediction. The experiments of 5‐fold Cross‐Validations (CV) yield mean accuracies of 88.96%, 86.37%, 82.88%, and 76.92% on four golden standard data sets (enzyme, ion channel, G protein-coupled receptors (GPCRs), and nuclear receptor, respectively). Moreover, the paper also conducts the state‐of‐art light gradient boosting machine (LGBM) and support vector machine (SVM) to further verify the performance of the proposed model. The experimental outcomes substantiate that the established model is feasible and reliable to predict DTIs. There is an excellent prospect that our model is capable of predicting DTIs as an efficient tool on a large scale.

Satellite networks face escalating cybersecurity threats from evolving attack vectors and systemic complexities. This paper proposes SatGuard, a novel framework integrating a three-dimensional penetration testing methodology and a nonlinear risk assessment mechanism tailored for satellite security. To address limitations of conventional tools in handling satellite-specific vulnerabilities, SatGuard employs large language models (LLMs) like GPT-4 and DeepSeek-R1. By leveraging their contextual reasoning and code-generation abilities, SatGuard enables semi-automated vulnerability analysis and exploitation. Validated in a simulated ground station environment, the framework achieved a 73.3% success rate (22/30 attempts) across critical ports, with an average of 5.5 human interactions per test. By bridging AI-driven automation with satellite-specific risk modeling, SatGuard advances cybersecurity for next-generation space infrastructure through scalable, ethically aligned solutions.