Timely prediction of a patient’s intention to exit the bed is critical for fall prevention but remains under-explored in real-world care facilities. Existing solutions rely on wearable sensors, cameras, pressure mats, or vibration sensors. However, these existing solutions may raise privacy concerns, incur high hardware costs, require extensive data annotation, and are sometimes noise-sensitive. These systems are typically validated in small laboratory studies or simulated environments, which lack the scale and diversity needed to assess real-world generalization. In this study, we propose a novel bed-exit intention prediction system based on a single load sensor installed under one bed leg and a Transformer-based prediction model, BEDFormer. BEDFormer introduces a Time‑aware De‑stationary Attention mechanism that embeds relative temporal information and dynamically adapts to the non-stationarity of load sensor signals over time. This prediction model can be trained with either annotated or quick heuristic labels for rapid deployment. We collected six months of bed-exit data from 95 beds in a care facility and demonstrated that BEDFormer outperforms recent state‑of‑the‑art time series baselines on all major metrics. Additionally, in a 19‑day field test at a senior care facility, the system achieved an average F1 score of 0.92 across nine residents, confirming strong cross‑site robustness. By turning an inexpensive, privacy‑preserving load sensor into a reliable early‑warning device, BEDFormer lowers the barrier to proactive fall prevention in healthcare settings.

Industrial asset management (IAM) is crucial for the sustainability and efficiency of industries that depend on significant infrastructure, especially pipelines. Pipelines are vital assets in the global energy supply chain, transporting oil, natural gas, and chemical products, and their maintenance is essential to prevent severe environmental and economic consequences. However, current methods such as manual visual inspections are limited by their invasiveness, the requirement for periodic shutdowns, and a lack of real-time accuracy. These limitations present substantial challenges to effective IAM. This paper introduces an innovative digital twin ecosystem integrated with information and communication technology to enhance IAM for pipelines. This ecosystem creates a dynamic, interactive digital twin that accurately reflects the physical state of pipelines, bolstered by real-time data transmission from sensors. The ecosystem comprises physical pipelines equipped with sensors, a comprehensive data knowledge library that records and updates damage information, a virtual pipeline twin, and an interactive platform that facilitates detailed visualization and interaction. Through a detailed case study of pipeline damage detection on cracks and corrosion using visual imaging techniques, this paper demonstrates enhanced results and visualizations compared to existing methods. The observed damage features sharper contrasts, higher resolution, and clearer boundaries in affected areas, significantly improving the accuracy of damage localization. Additionally, the relative accuracy of the calculated damage stress intensity factor by the virtual twin model reaches 95%. In summary, the capabilities for real-time, remote interaction and comprehensive visualization within the digital twin ecosystem significantly enhance the management efficiency of digital and intelligent pipeline IAM.

In medical image classification, uncertainty estimation providing confidence of decision is part of interpretability of prediction model. Based on estimated uncertainty, physicians can pick out cases with high uncertainty for further inspection. However, in this uncertainty-informed decision referral, models may make wrong predictions with high certainty which leads to omission of false predictions. Therefore, we propose a method to set up a model which could make correct prediction with low uncertainty and wrong prediction with high uncertainty. We integrate uncertainty estimation into training phase and design a novel loss function “penalized entropy” by penalizing wrong but certain samples to improve the models' certainty performance. Experiments were conducted on three datasets: optical coherence tomography (OCT) image dataset for anti-vascular endothelial growth factor (anti- VEGF) effectiveness classification, OCT image dataset for diagnostic classification, and chest X-ray dataset for pneumonia classification. Performances were evaluated on both accuracy metrics such as accuracy, sensitivity, specificity, area under the curve (AVC), and certainty metrics which are accuracy vs. uncertainty (AvV), probability of correct results among certain predictions (PCC), and probability of uncertain results among wrong predictions (PUW). Results show that the method using the proposed loss function can achieve better or comparable accuracy and state-of-the-art certainty performance.

Benefiting from low storage costs and high retrieval efficiency, hash learning has been a widely adopted technology for approximating nearest neighbor in large-scale data retrieval. Deep learning to hash greatly improves image retrieval performance by integrating feature learning and hash coding into an end-to-end framework. However, subject to application scope, most existing deep hashing methods only apply to retrieval of complete data and have undesirable results when retrieving incomplete but valuable data. In this paper we propose IDHashGAN, a novel deep hashing model with generative adversarial networks to retrieve incomplete data, in which feature restoration, feature learning and hash coding are integrated into an unified end-to-end framework. The proposed model consists of four key components: (1) reconstructive and generative loss are used to generate continuous feature of incomplete data in generative network; (2) supervised manifold similarity is proposed to improve retrieval accuracy and obtain good user acceptance; (3) adversarial and classified loss are designed to distinguish authenticity and similarity in discriminative network; and (4) encoding and quantization loss are adopted to preserve similarity and control hash quality. Extensive experiments on benchmark datasets show that IDHashGAN is competitive on complete dataset and yields substantial boosts of 70% on incomplete datasets compared to state-of-the-art hashing methods.

Considering the problems of poor effect, long reconstruction time, large mean square error (MSE), low signal‐to‐noise ratio (SNR), and structural similarity index (SSIM) of traditional methods in three‐dimensional (3D) image virtual reconstruction, the effect of 3D image virtual reconstruction based on visual communication is proposed. Using the distribution set of 3D image visual communication feature points, the feature point components of 3D image virtual reconstruction are obtained. By iterating the 3D image visual communication information, the features of 3D image virtual reconstruction in visual communication are decomposed, and the 3D image visual communication model is constructed. Based on the calculation of the difference of 3D image texture feature points, the spatial position relationship of 3D image feature points after virtual reconstruction is calculated to complete the texture mapping of 3D image. The deep texture feature points of 3D image are extracted. According to the description coefficient of 3D image virtual reconstruction in visual communication, the virtual reconstruction results of 3D image are constrained. The virtual reconstruction algorithm of 3D image is designed to realize the virtual reconstruction of 3D image. The results show that when the number of samples is 200, the virtual reconstruction time of this paper method is 2.1 s, and the system running time is 5 s; the SNR of the virtual reconstruction is 35.5 db. The MSE of 3D image virtual reconstruction is 3%, and the SSIM of virtual reconstruction is 1.38%, which shows that this paper method can effectively improve the ability of 3D image virtual reconstruction.

Person re-identification (Re-ID) aims at retrieving a person's identification across multiple non-overlapping cameras. While recent Re-ID methods have achieved significant success on a number of benchmark datasets, most of them are still insufficient in the scenes with unified or highly similar dressing like construction sites, schools and factories. To address this problem, we propose a supervisory mask attention network (SMA-Net). Our approach combines two key components: (1) ROI mask mapping (RMM) is a supervisory branch to provide ROI mappings that divide a person region into several parts; (2) Partial mask attention (PMA) integrates channel and space attention mechanisms that focus on local features and different accessories in each ROI. Therefore, the network can pay more attention to the local features and different accessories herein. Compared with the state of the art methods, SMA-Net demonstrates excellent performance on our dataset of construction scenes, with improvement of 6.65% in mAP and 3.8% in Rank-1 accuracy.

Large-scale knowledge graphs (KGs) play a critical role in question answering over KGs (KGs-QA). Despite of large scale, KGs suffer from incompleteness which results in the low-precision of KGs-QA. Since existing researches of relation prediction process each triple independently, the hidden relations which are inherently present in the neighborhood can not be captured. Complementarily, to capture both entity and relation features in a given entity's neighborhood, an entity importance estimation network of attention-based graph embedding is proposed, which consists of the attention-based graph embedding module and the entity importance estimation module. The new embedding of an entity from its n-hop neighbor is learned by an attention-based graph embedding module, and then the learned embedding is integrated into the entity importance estimation module to find entities of high importance in n-hop neighbors of the central entity. Multi-hop relations are encapsulated and an auxiliary edge of n-hop neighbors is introduced, which realizes the relation prediction task. In addition, we are the first to realize KGs-QA while realizing relation prediction, which alleviates the phenomenon of missing relations and the low-precision problem of KGs-QA. As expected, KGs-QA's F1 increases significantly when relation prediction is invoked. In 80% missing relation, we observe a F1 increase of 21, 17, 18 and 16 points over the no-relation prediction configuration for SQ, WQ, GQ and QALD-5, respectively.

In the literature of astronomical spectral classification, linear principle component analysis (PCA) was frequently employed to extract features of spectra data. However, the spectral data are too complicated to be well described by a linear model. In this paper, kernel independent component analysis (KICA), which contains a nonlinear kernel mapping component, is adopted to extract features from the spectra of galaxies. Then, a radial basis function neural network is adopted as a classifier to implement the classification. Experiments with real-world spectral data set show that KICA is a very appropriate technique to describe the important features of celestial objects, and the correct classification rate is improved compared with PCA method.

By analyzing the complex of the integration of heterogeneous database system and the characteristic of a multi-agent system, this paper proposes an agent based architecture that enables actual cooperation among a set of autonomous and heterogeneous information source. Three types of agent which is based on the speech-act theory have been defined in our proposal, coordinately, a new structure of the heterogeneous database system is also introduced.

In this paper, a novel stellar spectral classification technique is proposed. Which is composed of the following two steps: In the first step, Kalman filter is adopted to conduct de-noising process. At the same time, Kalman filter is also used for optimal feature extraction. The second step, radial basis function neural network is employed for the final classification. The proposed technique can be considered as a composite classifier which combines Kalman filter and radial basis function networks. The experiments show that our new technique is both robust and efficient, the obtained correct classification rate is much improved by the composite classifier, and these results are much better than the best results obtained from regularized discriminant analysis with principle component analysis data dimension reduction technique.