Missing values in time series data present a significant challenge, often degrading the performance of downstream tasks such as classification and forecasting. Traditional approaches address this issue by first imputing the missing values and then independently solving the predictive tasks. Recent methods have leveraged self-attention models to enhance imputation quality and accelerate inference. These models, however, predict values based on all input observations—including the missing values—thereby potentially compromising the fidelity of the imputed data. In this paper, we propose the Uncertainty-Aware Self-Attention (UASA) model to overcome these limitations. Our approach introduces two novel techniques: (i) A self-attention mechanism with a partially observed diagonal that effectively captures complex non-local dependencies in time series data—a characteristic also observed in fractional-order systems. This approach draws inspiration from fractional calculus, where non-integer-order derivatives better characterize complex dynamical systems with long-memory effects, providing a more comprehensive mathematical framework for handling temporal data. And (ii) uncertainty quantification in data imputation to better inform downstream tasks. The UASA model comprises an upstream component for data imputation and a downstream component for time series prediction, trained jointly in an end-to-end fashion to optimize both imputation accuracy and task-specific objectives simultaneously. For classification tasks, the UASA model demonstrates remarkable performance even under high missing data rates, achieving a ROC-AUC of $99.5\%$, a PR-AUC of $58.5\%$, and an F1-SCORE of $49.3\%$. For forecasting tasks on the AUST-Gait dataset, the UASA model achieves a Mean Squared Error (MSE) of 0.72 under $0\%$ missing data conditions (i.e., complete data input). Under the end-to-end training strategy evaluated across all missing data rates, the model achieves an average MSE of 0.74, showcasing its adaptability and robustness across diverse missing data scenarios.

This research paper outlines the development of a modular and adjustable transfer care robot aimed at enhancing safe and comfortable transfers for individuals with lower limb disabilities. To design this robot, we utilized a 3D motion capture system to analyze the movements of a person assisting another person and to determine the necessary range of motion and workspace for the robot. Based on this analysis, we developed a 3-UPS + UPR parallel spreader to transfer the person receiving care. We also conducted kinematic and dynamic analyses of the parallel spreader to validate its operational space and to obtain the force change curve for the drive. To evaluate the robot’s performance, we enlisted the help of ten volunteers with varying heights and weights. Our findings indicate that the pressure distribution during transfers remained largely consistent. Additionally, the surveys revealed that those receiving care perceived the robot as being capable of securely and comfortably transferring individuals between different assistive devices. This modular and adaptable transfer assistance robot presents a promising solution to the challenges encountered in caregiving.

Three-dimensional human pose estimation from monocular video remains challenging for clinical gait analysis due to high computational cost and the need for temporal consistency. We present Pose3DM, a bidirectional Mamba-based state-space framework that models intra-frame joint relations and inter-frame dynamics with linear computational complexity. Replacing transformer self-attention with state-space modeling improves efficiency without sacrificing accuracy. We further incorporate fractional-order total-variation regularization to capture long-range dependencies and memory effects, enhancing temporal and spatial coherence in gait dynamics. On Human3.6M, Pose3DM-L achieves 37.9 mm MPJPE under Protocol 1 (P1) and 32.1 mm P-MPJPE under Protocol 2 (P2), with 127 M MACs per frame and 30.8 G MACs in total. Relative to MotionBERT, P1 and P2 errors decrease by 3.3% and 2.4%, respectively, with 82.5% fewer parameters and 82.3% fewer MACs per frame. Compared with MotionAGFormer-L, Pose3DM-L improves P1 by 0.5 mm and P2 by 0.4 mm while using 60.6% less computation: 30.8 G vs. 78.3 G total MACs and 127 M vs. 322 M per frame. On AUST-VisGait across six gait patterns, Pose3DM consistently yields lower MPJPE, standard error, and maximum error, enabling reliable extraction of key gait parameters from monocular video. These results highlight state-space models as a cost-effective route to real-time gait assessment using a single RGB camera.

Food security serves as a vital cornerstone for social stability. As one of the most important staple crops globally, the quality and geographical origin of rice are directly associated with consumer health. Traditional methods for classifying rice by origin and variety rely on sensory evaluation and manual inspection, which are subject to uncertainty and human error. To address this, this paper proposes a method for classifying rice by origin and variety based on terahertz time-domain spectroscopy. Terahertz technology features the advantages of non-destructive, high-sensitivity and non-contact detection, making it well-suited for food detection. This study employs terahertz time-domain spectroscopy combined with machine learning modeling methods, using 20 types of rice as the subject of investigation, with a focus on modeling and analyzing four representative samples. Refractive index and absorption coefficient were extracted through preprocessing methods including Savitzky–Golay convolution smoothing, wavelet denoising and moving average smoothing. Modeling, classification, and detection were implemented using principal component analysis, partial least squares discriminant analysis, and least-squares support vector machine. The experimental results indicate that principal component analysis (PCA) alone performs poorly in classification tasks. However, a classification model combining PCA for dimensionality reduction with a least-squares support vector machine (SVM), following Savitzky–Golay smoothing, demonstrated the best performance, achieving a prediction accuracy of 93.3%. In an extended test involving 20 samples, the model achieved an identification accuracy of 89.6%. Quantitative metrics demonstrate the feasibility of using terahertz technology combined with optimized machine learning algorithms for classifying rice by origin and variety.

Aiming to address the problems of low accuracy in coal–rock identification during coal mining, which lead to energy waste and safety hazards, a high-precision coal–rock medium identification method combining terahertz time-domain spectroscopy technology and multiple machine learning algorithms is proposed. By preparing coal–rock samples with a gradient change in coal content, terahertz time-domain spectroscopy data of coal–rock mixed media are collected, and optical parameters such as the refractive index and absorption coefficient are extracted. Principal component analysis is used to reduce the dimensionality of the terahertz data, and machine learning algorithms such as support vector machine, least squares support vector machine, artificial neural networks, and random forests are adopted for classification and identification. The study found that terahertz waves are more sensitive to coal–rock media in the 0.7–1.3 THz frequency band, and that the refractive index and absorption coefficient of coal–rock mixed media are significantly positively correlated with coal content within the range of 0–30%. After feature extraction and K-fold cross-validation, the random forest model achieved a coal–rock classification accuracy of over 96% on the test set, significantly outperforming other comparison algorithms. The research verifies the efficiency and practicality of terahertz technology combined with multiple machine learning algorithms in coal–rock identification, providing a new method for fields such as mineral separation. This method has, to a certain extent, broken through the accuracy bottleneck of traditional coal–rock identification technologies within its applicable range, providing a new solution for real-time detection of coal–rock interfaces and is expected to further reduce the risks of ineffective mining and roof accidents in the future.

Modern turbine engines, when operating at high temperatures, can inhale calcium–magnesium–alumina–silicate particles (CaO-MgO-Al₂O₃-SiO₂, CMAS) from the air, which can erode the thermal barrier coatings on the blade surface, affecting the service life of the thermal barrier coatings and, in severe cases, leading to premature blade failure. Therefore, it is of great significance to effectively detect the thickness of CMAS deposited on the surface of the thermal barrier coatings at an early stage of CMAS erosion to ensure the high-temperature structural integrity of the hot-end components of aeroengines. Based on this, this study proposes a method combining terahertz time-domain spectroscopy technology and a hybrid machine learning algorithm for the quantitative detection of the thickness of CMAS on the surface of thermal barrier coatings. Firstly, the terahertz time-domain spectroscopy experimental data of CMAS were obtained using a terahertz experimental system, and the refractive index and absorption coefficient of CMAS in the terahertz frequency band were calculated. The FDTD method, Gaussian noise addition, and wavelet denoising processing were combined to further simulate the terahertz detection process of thermal barrier coatings with different thicknesses of CMAS attached to the surface under high-temperature conditions, and the terahertz simulation detection data were obtained. Principal component analysis (PCA) was used to reduce the dimensionality of the original experimental and simulation data, and a support vector machine (SVM) model integrating PCA and bacterial foraging optimization (BFO) algorithm was constructed. The research results show that the integrated model exhibits excellent performance in predicting the thickness of CMAS, with a correlation coefficient of 0.95, and the mean absolute error (MAE) and root mean square error (RMSE) are 0.13 μm and 0.46 μm, respectively. This study provides a new high-precision method for non-destructive detection of the thickness of CMAS on the surface of thermal barrier coatings, which has certain engineering application value for ensuring the service performance of thermal barrier coatings under harsh service conditions. Although the current method is based on simulated and experimental data under controlled conditions, it has the potential to be developed into an in situ monitoring strategy in the future, enabling real-time assessment of CMAS thickness on the coating surface during engine operation.