Logo: to the web site of Uppsala University

Road ponding, a prevalent traffic hazard, poses a serious threat to road safety by causing vehicles to lose control and leading to accidents ranging from minor fender benders to severe collisions. Existing technologies struggle to accurately identify road ponding due to complex road textures and variable ponding coloration influenced by reflection characteristics. To address this challenge, we propose a novel approach called Self-Attention-based Global Saliency-Enhanced Network (AGSENet) for proactive road ponding detection and traffic safety improvement. AGSENet incorporates saliency detection techniques through the Channel Saliency Information Focus (CSIF) and Spatial Saliency Information Enhancement (SSIE) modules. The CSIF module, integrated into the encoder, employs self-attention to highlight similar features by fusing spatial and channel information. The SSIE module, embedded in the decoder, refines edge features and reduces noise by leveraging correlations across different feature levels. To ensure accurate and reliable evaluation, we corrected significant mislabeling and missing annotations in the Puddle-1000 dataset. Additionally, we constructed the Foggy-Puddle and Night-Puddle datasets for road ponding detection in low-light and foggy conditions, respectively. Experimental results demonstrate that AGSENet outperforms existing methods, achieving IoU improvements of 2.03%, 0.62%, and 1.06% on the Puddle-1000, Foggy-Puddle, and Night-Puddle datasets, respectively, setting a new state-of-the-art in this field. Finally, we verified the algorithm's reliability on edge computing devices. This work provides a valuable reference for proactive warning research in road traffic safety. The source code and datasets are placed in the https://github.com/Lyu-Dakang/AGSENet.

The real-time contactless perception of human thermal comfort is essential for building energy conservation. Most of the traditional human thermal comfort perception methods are contact-type, which requires the use of sensors and other equipment to collect human physiological signals, and the human discomfort is often caused. The existing contactless methods are difficult to measure in real time and are not universal. An end-to-end noncontact human thermal comfort posture estimation algorithm is proposed in this paper, which focuses on solving the accuracy and real-time issues of non-contact human thermal comfort perception. This algorithm first trains a network to extract the heat map of human joints based on the ResNet structure, and then obtains the pose vector of the human body through the method of non-maximum suppression, and finally uses the LSTM network to process the extracted human pose vector to obtain the current person's posture actions and determine its current thermal comfort state based on posture actions. At the end of this paper, the effectiveness and robustness of the proposed algorithm are verified through test set verification, real-time detection and other methods. By integrating the algorithm into the building's air conditioning system, it continuously monitors the thermal comfort state of people within the area. Based on the estimated results of thermal comfort postures, it dynamically optimizes the operation settings of the air conditioning system, enhancing personnel comfort while reducing energy waste.

This article focuses on optimizing the elasticity of spacecraft by harnessing the power of artificial intelligence (AI) technology. With the support of spatial informatics and digital twins technology, this work initially employs AI techniques, specifically the radial basis function neural networks, a deep learning algorithm, to perform global optimization and orbit fitting for spacecraft. Augmented Lagrangian multipliers are then introduced to locally optimize this neural network. Additionally, to further enhance the spacecraft's flexibility, an improved particle swarm optimization (PSO) algorithm is applied to optimize the proposed network. The work also introduces a periodic variational multiobjective quantum particle swarm optimization (PMQPSO) algorithm. Subsequently, a rigid-flexible coupled dynamics model for the spacecraft is established, and relevant simulations and experiments are conducted to support this work. The results indicate that the average fitness of the improved PMQPSO algorithm decreases to 18.23 after 500 iterations, with its performance being at least 3.2% higher than that of the classical quantum PSO algorithm. Furthermore, after the initial decline in the first order, the limiter residuals no longer decline and exhibit convergence, as the residual curve transitions from high to low, indicating a gradual improvement in convergence and stability. These findings highlight the advantages of the PMQPSO algorithm in optimizing the spacecraft's elasticity. In conclusion, this parameter optimization holds practical significance for the design optimization of aircraft aerodynamic shapes.

To date, Parkinson's disease (PD) is an incurable neurological disorder, and the time of quality life can only be extended through early detection and timely intervention. However, the symptoms of early PD are both heterogeneous and subtle. To cope with these challenges, we develop a two-level fusion framework for smart healthcare, leveraging smartphones interconnected with the Internet of Medical Things and exploring the contribution of multi-sensor and multi-activity data. Rotation rate and acceleration during walking activity are recorded with the gyroscope and accelerometer, while location coordinates and acceleration during tapping activity are collected via the touch screen and accelerometer, and voice signals are captured by the microphone. The main scientific contribution is the enhanced fusion of multi-sensor information to cope with the heterogeneous and subtle nature of early PD symptoms, achieved by a first-level component that fuses features within a single activity using an attention mechanism and a second-level component that dynamically allocates weights across activities. Compared with related works, the proposed framework explores the potential of fusing multi-sensor data within a single activity, and mines the importance of different activities that correspond to early PD symptoms. The proposed two-level fusion framework achieves an AUC of 0.891 (95 % CI, 0.860-0.921) and a sensitivity of 0.950 (95 % CI, 0.888-1.000) in early PD detection, demonstrating that it efficiently fuses information from different sensor data for various activities and has a strong fault tolerance for data.

In the current realms of scientific research and commercial applications, the risk inference of regulatory violations by publicly listed enterprises has attracted considerable attention. However, there are some problems in the existing research on the deduction and prediction of violation risk of listed enterprises, such as the lack of analysis of the causal logic association between violation events, the low interpretability and effectiveness of the deduction and the lack of training data. To solve these problems, we propose a framework for enterprise violation risk deduction based on generative AI and event evolution graphs. First, the generative AI technology was used to generate a new text summary of the lengthy and complex enterprise violation announcement to realize a concise overview of the violation matters. Second, by fine-tuning the generative AI model, an event entity and causality extraction framework based on automated data augmentation are proposed, and the UIE (Unified Structure Generation for Universal Information Extraction) event entity extraction model is used to create the event entity extraction for listed enterprises 'violations. Then, a causality extraction model CDDP-GAT (Event Causality Extraction Based on Chinese Dictionary and Dependency Parsing of GAT) is proposed. This model aims to identify and analyse the causal links between corporate breaches, thereby deepening the understanding of the event logic. Then, the merger of similar events was realized, and the causal correlation weights between enterprise violation-related events were evaluated. Finally, the listed enterprise's violation risk event evolution graph was constructed, and the enterprise violation risk deduction was carried out to form an expert system of financial violations. The deduction results show that the method can effectively reveal signs of enterprise violations and adverse consequences.

Vehicle edge computing (VEC) can effectively meet the demand for computing resources in autonomous driving. However, complex resource constraints exist in the practical application of VEC, making offloading tasks a key challenge. Traditional scheduling algorithms are usually optimized only for latency and cost and can handle only a small number of tasks; however, they cannot handle real-world intensive vehicle-road cooperation scenarios involving many tasks. Thus, this article constructs a large-scale multiobjective computing offloading optimization model that comprehensively considers latency, energy consumption, load balancing, and resource utilization. To improve the offloading performance of VEC, we propose a large-scale multiobjective optimization algorithm with hybrid directed sampling and adaptive offspring generation (LMOEA-HDGS). The algorithm can generate adaptive offspring by sampling in two types of search directions in the decision space and can adapt to the complex shape of the Pareto front while balancing diversity and convergence. The experimental results show that the proposed algorithm can effectively optimize the task offloading problem of VEC in an intensive vehicle-road cooperation scenario.

Deep learning has made remarkable achievements in many fields. However, although fuzzy neural networks with natural interpretability are widely used in prediction and control scenarios, there are very few studies on the deepening of fuzzy systems. By integrating rough set theory, fuzzy rough neural network has unique advantages benefiting from the complementarity of the fuzzy set theory and the rough set theory as well as the powerful learning ability of neural networks. Similarly, research on deep fuzzy rough neural networks is even rarer. In this article, in order to improve the performance of the fuzzy rough neural network and expand its application range, the deep fuzzy rough neural network model is constructed and optimized by stacking blocks of fuzzy rough neural network to imitate the deepening deep neural network based on multiobjective evolution. Each fuzzy rough neural network block is interpretable, and its stacked architecture also has high interpretability. To automatically generate deep fuzzy rough neural network models with high efficiency, a distributed parallel multiobjective neuroevolution framework is developed, thus blocks can be stacked flexibly and deep architecture can be optimized considering multiple optimization objectives of accuracy, interpretability, and generalization simultaneously. In addition, multiobjective evolution is combined with Wang-Mendel method, pseudoinverse, and backpropagation to effectively learn specific parameters. Finally, based on the time series prediction problems, the superiority of the multiobjective deep fuzzy rough neural network evolutionary framework is verified.

Smoke detection is essential for fire prevention, yet it is significantly hampered by the visual similarities between smoke and fog. To address this challenge, a split top-k attention transformer framework (STKformer) is proposed. The STKformer incorporates split top-k attention (STKA), which partitions the attention map for top-k selection to retain informative self-attention values while capturing long-range dependencies. This approach effectively filters out irrelevant attention scores, preventing information loss. Furthermore, the adaptive dark-channel-prior guidance network (ADGN) is designed to enhance smoke recognition under foggy conditions. ADGN employs pooling operations instead of minimum value filtering, allowing for efficient dark channel extraction with learnable parameters and adaptively reducing the impact of fog. The extracted prior information subsequently guides feature extraction through a priorformer block, improving model robustness. Additionally, a cross-stage fusion module (CSFM) is introduced to aggregate features from different stages efficiently, enabling flexible adaptation to smoke features at various scales and enhancing detection accuracy. Comprehensive experiments demonstrate that the proposed method achieves state-of-the-art performance across multiple datasets, with an accuracy of 89.68% on dataset for smoke detection in fog, 99.76% on CCTV images of smoke, and 99.76% on UAV images of wildfire. The method maintains high speed and lightweight characteristics, validated with an inference speed of 211.46 FPS on an NVIDIA Jetson AGX Orin after TensorRT acceleration, confirming its effectiveness and efficiency for real-world applications. The source code is available at https://github.com/Jiongze-Yu/STKformerhttps://github.com/Jiongze-Yu/STKformer.

The regulatory compliance supervision of listed enterprises is of great significance for ensuring the stable operation of financial markets. However, existing graph propagation algorithms used to identify corporate violations are limited by their inherent randomness. At the same time, current methods consider a relatively narrow range of risk dimensions, making it difficult to accurately distinguish the different characteristics of enterprises. To this end, this paper designs a Propagation of Violation Risks for Listed Enterprises (PVR-LE) algorithm, which reduces the risk of false propagation through a risk weight decay mechanism and dynamic updating of risk propagation patterns. Furthermore, a Multi-source Risk Fusion Neural Network (MRFNN) for corporate violation prediction tasks is proposed, which fuses the liquidity risk characteristics between enterprises, clustering characteristics, and the enterprise's risk characteristics to endow the violating enterprise nodes with more distinctive and characteristic vector features, thereby identifying whether there are violations in the company. At the same time, a generative adversarial network is used to generate samples of violating enterprises to solve the problem of class imbalance. Experiments are conducted on areal dataset constructed from information on Chinese listed companies, and this method improves the accuracy, recall, the weighted harmonic mean of precision and recall(F1-score), and geometric mean(G-mean) metrics by 2.12%, 3.19%, 2.14%, and 2.79%, respectively, compared to the best performance of existing models. The experimental results show that the proposed method effectively improves the accuracy of identifying corporate violations and helps promote the development of intelligent regulatory work for listed enterprises.

Smartphones are now being further developed for healthcare monitoring. Owing to the ubiquity of smartphones, it lays the foundation for widespread promotion and can contribute to promoting medical equity. Relying on its portability, continuity, and realtime, it is possible to continuously monitor relevant digital biomarkers anytime, anywhere. Currently, some advanced and interesting works based on smartphones have emerged. This paper surveys the current state of the art in the fast-moving field of smartphones for measurement of cardiovascular healthcare. We systematically explore the capabilities of smartphone-embedded sensors for cardiovascular healthcare and analyze the signals acquired by smartphones. We also summarize the applications of smartphone-based cardiovascular healthcare: cardiac parameter measurement (heart rate, blood oxygen, blood pressure, and blood glucose), and cardiovascular monitoring applications (daily monitoring, healthy vascular aging detection, and disease screening). Smartphone-based disease screening currently only includes some simple cardiovascular diseases. Finally, we discuss the challenges faced in the research.