Jinan Sun

This paper presents a wolf-pack-inspired distributed encirclement framework for heterogeneous unmanned aerial vehicle–unmanned surface vehicle (UAV–USV) teams operating with limited communication and intermittent visual sensing. The proposed growth-based visual connectivity encirclement (GB-VCE) method separates high-level role evolution from low-level motion execution. Each agent accumulates local evidence from target observation, visual connectivity contribution, motion stability, and platform characteristics, enabling smooth transitions among leader, relay, searcher, and encircler roles. Visual links and relay relations are treated as coordination resources rather than auxiliary sensing constraints, allowing target-related information to propagate through the team without centralized fusion or persistent all-to-all communication. Simulations with paired random seeds show that GB-VCE improves encirclement accuracy, target-information coverage, convergence consistency, and heterogeneous role complementarity compared with fixed-role, static-leader, UAV-only, and USV-only baselines. The results indicate that biomimetic role growth and visual connectivity preservation provide an interpretable and robust coordination principle for air–sea cooperative encirclement.

Facial palsy, affecting 1 in 60 individuals, requires precise assessment for effective treatment and follow-up. Current grading systems are inconsistent due to redundant subjective factors. Discrete assessment of the whole face fails to address eyelid and mouth movements innervated by separate nerves. Herein, we proposed modified House-Brackmann (H-B) criteria with continuous and componentized measurements for the eyelids and mouth. Using a dataset of 274 patients with vestibular schwannoma (VS), we developed a dense facial landmark alignment model. This model outperformed existing algorithms in facial palsy detection and produced clinically interpretable asymmetric coefficients of eyelids and mouth correlated with consensus H-B grades (r = 0.892 and 0.890, P < 0.001). Thresholds from ROC analysis transformed these coefficients into grades, comparable to consensus H-B grades. Validated in a separate post-VS multi-center cohort, this model showed high accuracy (eyelid Grade 6 not observed; analyses restricted accordingly) and potential for continuous monitoring of facial palsy, offering a more precise and comprehensive assessment tool. Trial Registration Information the Chinese Clinical Trial Registry (ChiCTR1900024638; July 19, 2019).

In the domain of multiplayer pursuit–evasion games, it is crucial to address the practical aspects of the players' heterogeneity, the distributed control manner, and the pursuers' goal of minimum makespan. However, the three topics have received limited attention in existing literature, both separately and in combination. In this article, we address the multiplayer pursuit–evasion game integrating these key topics, where the pursuers with simple motions strive to capture as many evaders, characterized by damped double integrators, as possible, meanwhile minimizing the task makespan. To this end, we establish an overall framework and sequentially tackle four key issues, leading to an effective solution to the entire problem. We first propose a novel isochron-based method to derive the capture condition and acquire the optimal pursuit and evasion strategies. Then, we enhance the usability of the capture condition by geometrically deriving the analytical form of the pursuer's winning region with respect to a given evader. Third, leveraging the concept of winning regions, we determine the lower boundary of pursuers' sensing range, ensuring sufficient information acquisition for evader allocation while avoiding selection conflicts. Finally, we propose a fully distributed allocation algorithm for each pursuer allowing the pursuer team to converge to the optimal evader allocation. By combining these contributions, we successfully provide an effective solution to the entire problem. Various simulations are conducted to show the effectiveness of our proposed methods.

This paper explores the recent advancements in enhancing Computational Fluid Dynamics (CFD) tasks through Machine Learning (ML) techniques. We begin by introducing fundamental concepts, traditional methods, and benchmark datasets, then examine the various roles ML plays in improving CFD. The literature systematically reviews papers in recent five years and introduces a novel classification for forward modeling: Data-driven Surrogates, Physics-Informed Surrogates, and ML-assisted Numerical Solutions. Furthermore, we also review the latest ML methods in inverse design and control, offering a novel classification and providing an in-depth discussion. Then we highlight real-world applications of ML for CFD in critical scientific and engineering disciplines, including aerodynamics, combustion, atmosphere & ocean science, biology fluid, plasma, symbolic regression, and reduced order modeling. Besides, we identify key challenges and advocate for future research directions to address these challenges, such as multi-scale representation, physical knowledge encoding, scientific foundation model and automatic scientific discovery. This review serves as a guide for the rapidly expanding ML for CFD community, aiming to inspire insights for future advancements. We draw the conclusion that ML is poised to significantly transform CFD research by enhancing simulation accuracy, reducing computational time, and enabling more complex analyses of fluid dynamics. The paper resources can be viewed at https://github.com/WillDreamer/Awesome-AI4CFD.

As an important problem in searching system development, domain adaptive retrieval seeks to train a retrieval model with both labeled source samples and unlabeled target samples. Although several domain adaptive hashing algorithms have been proposed to handle the problem with high efficiency, they often presume that source and target domains share all classes. However, prior knowledge about the label space on the target domain is hard to obtain in reality. To tackle this, we study a novel and challenging problem of universal domain adaptive retrieval, which evidently increases the difficulty of effective domain alignment. In this paper, we propose a hashing method named Relational and prOtotypical Structure lEarning (ROSE) to solve the problem. In particular, to overcome domain shift, we construct a relational structure depicting cross-domain similar pairs based on ranking statistics, then learn from the structure by maximizing the similarity between similar pairs compared with challenging negatives. Moreover, target private samples are detected using the min-max criterion, which helps to construct hashing prototypes in the Hamming space. On this basis, we combine prototypical structure learning with online clustering in the Hamming space, which improves target semantic learning under label deficiency. Extensive experiments on several benchmarks demonstrate that our proposed ROSE significantly outperforms a wide range of state-of-the-art methods. Our source code is available at https://github.com/WillDreamer/Rose.git.

This paper aims to solve the multi-objective operating planning problem in the radioactive environment. First, a more complicated radiation dose model is constructed, considering difficulty levels at each operating point. Based on this model, the multi-objective operating planning problem is converted to a variant traveling salesman problem (VTSP). Second, with respect to this issue, a novel combinatorial algorithm framework, namely hyper-parameter adaptive genetic algorithm (HPAGA), integrating bio-inspired optimization with reinforcement learning, is proposed, which allows for adaptive adjustment of the hyperparameters of GA so as to obtain optimal solutions efficiently. Third, comparative studies demonstrate the superior performance of the proposed HPAGA against classical evolutionary algorithms for various TSP instances. Additionally, a case study in the simulated radioactive environment implies the potential application of HPAGA in the future.

Hashing aims to compress raw data into compact binary descriptors, which has drawn increasing interest for efficient large-scale image retrieval. Current deep hashing often employs evaluation protocols where usually query data and training data are from similar distributions. However, more realistic evaluations should take into account a broad spectrum of distribution shifts with varying degrees. Therefore, we study the problem of out-of-distribution generalization in image retrieval, which seeks to learn a retrieval model from a source domain and generalize to unseen target domains. However, this problem is challenging owing to data scarcity in target domains and the potential overfitting of domain-specific patterns. Here, we propose a novel hashing model named Looking-into-gradients (LOG) for image retrieval under out-of-distribution shifts, which comprehensively explores gradients for both data generation and model optimization. Specifically, to overcome data deficiency in target domains, we formalize the worst-case problem to generate challenging virtue samples via adversarial gradient ascend. Besides, to further enhance model generalization capability, we not only identify non-crucial parameters with minor gradients and values and shrink them to zero, but also modify the inconsistent gradients across domains to prevent learning domain-specific patterns. Extensive experiments on various datasets demonstrate that LOG outperforms state-of-the-art methods by up to 8.54%.

Due to the excellent computing efficiency, learning to hash has acquired broad popularity for Big Data retrieval. Although supervised hashing methods have achieved promising performance recently, they presume that all training samples are appropriately annotated. Unfortunately, label noise is ubiquitous owing to erroneous annotations in real-world applications, which could seriously deteriorate the retrieval performance due to imprecise supervised guidance and severe memorization of noisy data. Here we propose a comprehensive method DIOR to handle the difficulties of learning to hash with label noise. DIOR performs partitions from two complementary levels, namely sample level and parameter level. On the one hand, DIOR divides the dataset into a labeled set with clean samples and an unlabeled set with noisy samples using an ensemble of perturbed views. Then we train the network in a contrastive semi-supervised manner by reconstructing label embeddings for both reliable supervision of clean data and sufficient exploration of noisy data. On the other hand, inspired by recent pruning techniques, DIOR divides the parameters in the hashing network into crucial parameters and non-crucial parameters, and then optimizes them separately to reduce the overfitting of noisy data. Extensive experiments on four popular benchmark datasets demonstrate the effectiveness of DIOR.