Distributed Internet of Things (IoT) nodes powered by wind energy harvesters (WEHs) have enabled long-term, maintenance-free environmental monitoring in remote and extreme scenarios. However, the large fluctuations in wind speed may cause structural fatigue and dynamical instability of conventional WEHs. To achieve stable and durable operation under high wind speeds (up to 25 m/s), this study reports a Feynman-sprinkler-like WEH (FSL-WEH). In contrast to conventional WEHs that rely on turbine blades directly driven by the wind, the FSL-WEH harnesses the reaction force generated by airflow expelled through Feynman nozzles to produce rotational motion. Simulation and experiments indicate that the FSL-WEH achieves a maximum power output of 4.34 W and a peak power density of 66.77 kW/m3at a wind speed of 25 m/s. Moreover, the FSL-WEH successfully powers an environmental monitoring system, which wirelessly transmits temperature and humidity data to connected terminals, validating the operational reliability and durability of the device in extremely high wind speeds and high-humidity conditions. This study provides a promising solution for scavenging wind energy to power distributed environmental sensor networks under high-wind, high-humidity conditions.

Model fusion combines multiple Large Language Models (LLMs) with different strengths into a more powerful, integrated model through lightweight training methods. Existing works on model fusion focus primarily on supervised fine-tuning (SFT), leaving preference alignment (PA) —a critical phase for enhancing LLM performance—largely unexplored. The current few fusion methods on PA phase, like WRPO, simplify the process by utilizing only response outputs from source models while discarding their probability information. To address this limitation, we propose InfiFPO, a preference optimization method for implicit model fusion. InfiFPO replaces the reference model in Direct Preference Optimization (DPO) with a fused source model that synthesizes multi-source probabilities at the sequence level, circumventing complex vocabulary alignment challenges in previous works and meanwhile maintaining the probability information. By introducing probability clipping and max-margin fusion strategies, InfiFPO enables the pivot model to align with human preferences while effectively distilling knowledge from source models. Comprehensive experiments on 11 widely-used benchmarks demonstrate that InfiFPO consistently outperforms existing model fusion and preference optimization methods. When using Phi-4 as the pivot model, InfiFPO improves its average performance from 79.95 to 83.33 on 11 benchmarks, significantly improving its capabilities in mathematics, coding, and reasoning tasks.

Recent advances in large language models (LLMs) have intensified efforts to fuse heterogeneous open-source models into a unified system that inherits their complementary strengths. Existing logit-based fusion methods maintain inference efficiency but treat vocabulary dimensions independently, overlooking semantic dependencies encoded by cross-dimension interactions. These dependencies reflect how token types interact under a model's internal reasoning and are essential for aligning models with diverse generation behaviors. To explicitly model these dependencies, we propose \textbf{InfiGFusion}, the first structure-aware fusion framework with a novel \textit{Graph-on-Logits Distillation} (GLD) loss. Specifically, we retain the top-$k$ logits per output and aggregate their outer products across sequence positions to form a global co-activation graph, where nodes represent vocabulary channels and edges quantify their joint activations. To ensure scalability and efficiency, we design a sorting-based closed-form approximation that reduces the original $O(n^4)$ cost of Gromov-Wasserstein distance to $O(n \log n)$, with provable approximation guarantees. Experiments across multiple fusion settings show that GLD consistently improves fusion quality and stability. InfiGFusion outperforms SOTA models and fusion baselines across 11 benchmarks spanning reasoning, coding, and mathematics. It shows particular strength in complex reasoning tasks, with +35.6 improvement on Multistep Arithmetic and +37.06 on Causal Judgement over SFT, demonstrating superior multi-step and relational inference.

Multi-object detection in autonomous driving faces challenges due to multi-scale entities, diverse streetscapes, and limited computational resources. To address these challenges, we present MT-DyNN, a Multi-Teacher knowledge-distilled Dynamic Neural Network framework for instance-adaptive detection, optimizing detection accuracy and inference cost in autonomous driving. The framework’s student network comprises a customizable multi-branch residual detection network and a lightweight policy network. The former efficiently extracts multi-scale features in parallel without altering receptive fields, while the latter, depending on curriculum learning, captures task-relevant features and dynamically generates routing vectors to guide the activation or deactivation of residual blocks according to image instance complexity. The framework’s teacher network employs a soft-voting strategy to consolidate knowledge from multiple pre-trained teacher models, providing consistent guidance to the student. Within this distillation paradigm, the policy network’s routing search space is gradually refined, and the policy and detection networks are jointly fine-tuned to optimize the alignment between routing decisions and feature extraction. Experimental results on CIFAR and ImageNet demonstrate that compared to early exiting and stochastic depth methods, MT-DyNN achieves higher accuracy at the same inference cost and reduces the cost by 50% and 59% at comparable accuracy levels. The generated routing maintains channel sparsity across diverse scenarios.

Weakly Supervised Entity Alignment (EA) is the task of identifying equivalent entities across diverse knowledge graphs (KGs) using only a limited number of seed alignments. Despite substantial advances in aggregation-based weakly supervised EA, the underlying mechanisms in this setting remain unexplored. In this article, we present a propagation perspective to analyze weakly supervised EA and explain the existing aggregation-based EA models. Our theoretical analysis reveals that these models essentially seek propagation operators for pairwise entity similarities. We further prove that, despite the structural heterogeneity across different KGs, the potentially aligned entities within aggregation-based EA models exhibit isomorphic subgraphs, a fundamental yet underexplored premise of EA. Leveraging this insight, we introduce a potential isomorphism propagation operator to enhance the propagation of neighborhood information across KGs. We develop a general EA framework, PipEA, incorporating this operator to improve the accuracy of every type of aggregation-based model without altering the learning process. Extensive experiments substantiate our theoretical findings and demonstrate PipEA’s significant performance gains over state-of-the-art weakly supervised EA methods. Our work advances the field and enhances our comprehension of aggregation-based weakly supervised EA.

Multi-Modal Entity Alignment (MMEA) is a pivotal task in Multi-Modal Knowledge Graphs (MMKGs), seeking to identify identical entities by leveraging associated modal attributes. However, real-world MMKGs confront the challenges of semantic inconsistency arising from diverse and incomplete data sources. This inconsistency is predominantly caused by the absence of specific modal attributes, manifesting in two distinct forms: disparities in attribute counts or the absence of certain modalities. Current methods address these issues through attribute interpolation, but their reliance on predefined distributions introduces modality noise, compromising original semantic information. Furthermore, the absence of a generalizable theoretical principle hampers progress towards achieving semantic consistency. In this work, we propose a generalizable theoretical principle by examining semantic consistency from the perspective of Dirichlet energy. Our research reveals that, in the presence of semantic inconsistency, models tend to overfit to modality noise, leading to over-smoothing and performance oscillations or declines, particularly in scenarios with a high rate of missing modality. To overcome these challenges, we propose DESAlign, a robust method addressing the over-smoothing caused by semantic inconsistency and interpolating missing semantics using existing modalities. Specifically, we devise a training strategy for multi-modal knowledge graph learning based on our proposed principle. Then, we introduce a propagation strategy that utilizes existing features to provide interpolation solutions for missing semantic features. DESAlign outperforms existing approaches across 60 benchmark splits, encompassing both monolingual and bilingual scenarios, achieving state-of-the-art performance. Experiments on splits with high missing modal attributes demonstrate its effectiveness, providing a robust MMEA solution to semantic inconsistency in real-world MMKGs.

Anomaly detection in multivariate time series (MTS) is crucial for various applications in data mining and industry. Current industrial methods typically approach anomaly detection as an unsupervised learning task, aiming to identify deviations by estimating the normal distribution in noisy, label-free datasets. These methods increasingly incorporate interdependencies between channels through graph structures to enhance accuracy. However, the role of interdependencies is more critical than previously understood, as shifts in interdependencies between MTS channels from normal to anomalous data are significant. This observation suggests that anomalies could be detected by changes in these interdependency graph series. To capitalize on this insight, we introduce MADGA (MTS Anomaly Detection via Graph Alignment), which redefines anomaly detection as a graph alignment (GA) problem that explicitly utilizes interdependencies for anomaly detection. MADGA dynamically transforms subsequences into graphs to capture the evolving interdependencies, and Graph alignment is performed between these graphs, optimizing an alignment plan that minimizes cost, effectively minimizing the distance for normal data and maximizing it for anomalous data. Uniquely, our GA approach involves explicit alignment of both nodes and edges, employing Wasserstein distance for nodes and Gromov-Wasserstein distance for edges. To our knowledge, this is the first application of GA to MTS anomaly detection that explicitly leverages interdependency for this purpose. Extensive experiments on diverse real-world datasets validate the effectiveness of MADGA, demonstrating its capability to detect anomalies and differentiate interdependencies, consistently achieving state-of-the-art across various scenarios.

Multi-modal Entity Alignment (MMEA) refers to utilizing multiple modalities such as text, images, videos, etc., to match entities from multiple knowledge graphs. Compared to single-modal entity alignment, multi-modal entity alignment can provide a more comprehensive description of entity semantics and improve matching accuracy. Currently, research efforts are directed towards the development of sophisticated deep learning models, such as graph neural networks, that can effectively capture and integrate the multi-modal features of entities for entity alignment tasks. While these models have shown promising results, they tend to focus on capturing only the local structure of entities, leading to the challenge of subgraph isomorphism. Moreover, the complexity of these models often hinders their scalability. To address these limitations, this paper proposes a non-neural, position-enhanced multi-modal entity alignment algorithm that leverages the label propagation technique to fuse and aggregate various multi-modal and position features, resulting in entity representations that are aware of long-term alignment information. Extensive experiments on various public datasets demonstrate that our proposed approach outperforms state-of-the-art algorithms in terms of both alignment accuracy and computational efficiency.

Abstract Crowd evacuation has gained increasing attention in recent years. The agent‐based method has shown a superior capability to simulate complex behaviors during crowd evacuation simulation. For agent modeling, most existing methods only consider the decision process but ignore the detailed physical motion. In this article, we propose a hierarchical framework for crowd evacuation simulation, which combines the agent decision model with the agent motion model. In the decision model, we integrate emotional contagion and scene information to determine global path planning and local collision avoidance. In the motion model, we introduce a physics‐based character control method and control agent motion using deep reinforcement learning. Based on the decision strategy, the decision model can use a signal to control the agent motion in the motion model. Compared with existing methods, our framework can simulate physical interactions between agents and the environment. The results of the crowd evacuation simulation demonstrate that our framework can simulate crowd evacuation with physical fidelity.

Microsurgery involves the dexterous manipulation of delicate tissue or fragile structures, such as small blood vessels and nerves, under a microscope. To address the limitations of imprecise manipulation of human hands, robotic systems have been developed to assist surgeons in performing complex microsurgical tasks with greater precision and safety. However, the steep learning curve for robot-assisted microsurgery (RAMS) and the shortage of well-trained surgeons pose significant challenges to the widespread adoption of RAMS. Therefore, the development of a versatile training system for RAMS is necessary, which can bring tangible benefits to both surgeons and patients. In this paper, we present a Tactile Internet-Based Micromanipulation System (TIMS) based on a ROS-Django web-based architecture for microsurgical training. This system can provide tactile feedback to operators via a wearable tactile display (WTD), while real-time data is transmitted through the internet via a ROS-Django framework. In addition, TIMS integrates haptic guidance to ‘guide’ the trainees to follow a desired trajectory provided by expert surgeons. Learning from demonstration based on Gaussian Process Regression (GPR) was used to generate the desired trajectory. We conducted user studies to verify the effectiveness of our proposed TIMS, comparing users' performance with and without tactile feedback and/or haptic guidance. For more details of this project, please view our website: https://sites.google.com/view/viewtims/home.