Search

Multivariate time series modeling has been essential in sensor-based data mining tasks. However, capturing complex dynamics caused by intra-variable (temporal) and inter-variable (spatial) relationships while simultaneously taking into account evolving data distributions is a non-trivial task, which faces accumulated computational overhead and multiple temporal patterns or distribution modes. Most existing methods focus on the former direction without adaptive task-specific learning ability. To this end, we developed a holistic spatial-temporal meta-learning probabilistic inference framework, entitled ST-MeLaPI, for the efficient and versatile learning of complex dynamics. Specifically, first, a multivariate relationship recognition module is utilized to learn task-specific inter-variable dependencies. Then, a multiview meta-learning and probabilistic inference strategy was designed to learn shared parameters while enabling the fast and flexible learning of task-specific parameters for different batches. At the core are spatial dependency-oriented and temporal pattern-oriented meta-learning approximate probabilistic inference modules, which can quickly adapt to changing environments via stochastic neurons at each timestamp. Finally, a gated aggregation scheme is leveraged to realize appropriate information selection for the generative style prediction. We benchmarked our approach against state-of-the-art methods with real-world data. The experimental results demonstrate the superiority of our approach over the baselines. [ABSTRACT FROM AUTHOR]

EEG signals capture information through multi-channel electrodes and hold promising prospects for human emotion recognition. However, the presence of high levels of noise and the diverse nature of EEG signals pose significant challenges, leading to potential overfitting issues that further complicate the extraction of meaningful information. To address this issue, we propose a Granger causal-based spatial–temporal contrastive learning framework, which significantly enhances the ability to capture EEG signal information by modeling rich spatial–temporal relationships. Specifically, in the spatial dimension, we employ a sampling strategy to select positive sample pairs from individuals watching the same video. Subsequently, a Granger causality test is utilized to enhance graph data and construct potential causality for each channel. Finally, a residual graph convolutional neural network is employed to extract features from EEG signals and compute spatial contrast loss. In the temporal dimension, we first apply a frequency domain noise reduction module for data enhancement on each time series. Then, we introduce the Granger–Former model to capture time domain representation and calculate the time contrast loss. We conduct extensive experiments on two publicly available sentiment recognition datasets (DEAP and SEED), achieving 1.65% improvement of the DEAP dataset and 1.55% improvement of the SEED dataset compared to state-of-the-art unsupervised models. Our method outperforms benchmark methods in terms of prediction accuracy as well as interpretability. [ABSTRACT FROM AUTHOR]

In the subject recognition (SR) task under Knowledge Base Question Answering (KBQA), a common method is by training and employing a general flat Named-Entity Recognition (NER) model. However, it is not effective and robust enough in the case that the recognized entity could not be strictly matched to any subjects in the Knowledge Base (KB). Compared to flat NER models, nested NER models show more flexibility and robustness in general NER tasks, whereas it is difficult to employ a nested NER model directly in an SR task. In this paper, we take advantage of features of a nested NER model and propose an Improved Nested NER Model (INNM) for the SR task under KBQA. In our model, each question token is labeled as either an entity token, a start token, or an end token by a modified nested NER model based on semantics. Then, entity candidates would be generated based on such labels, and an approximate matching strategy is employed to score all subjects in the KB based on string similarity to find the best-matched subject. Experimental results show that our model is effective and robust to both single-relation questions and complex questions, which outperforms the baseline flat NER model by a margin of 3.3% accuracy on the SimpleQuestions dataset and a margin of 11.0% accuracy on the WebQuestionsSP dataset. [ABSTRACT FROM AUTHOR]

Abscisic acid (ABA) plays a crucial role in regulating the ripening of non-climacteric strawberry fruit. In the present study, ABA was confirmed to promote strawberry ripening and induce the down-regulation of FaMADS1. The transient silence of FaMADS1 in strawberries promoted fruit ripening and induced the content of anthocyanin and soluble pectin but reduced firmness and protopectin through a tobacco rattle virus-induced gene silencing technique. In parallel with the accelerated ripening, the genes were significantly induced in the transiently modified fruit, including anthocyanin-related PAL6, C4H, 4CL, DFR, and UFGT, softening-related PL and XTH, and aroma-related QR and AAT2. In addition, the interaction between FaMADS1 and ABA-related transcription factors was researched. Yeast one-hybrid analysis indicated that the FaMADS1 promoter could interact with FaABI5-5, FaTRAB1, and FaABI5. Furthermore, dual-luciferase assay suggested that FaTRAB1 could actively bind with the FaMADS1 promoter, resulting in the decreased expression of FaMADS1. In brief, these results suggest that the ABA-dependent ripening of strawberry fruit was probably inhibited through inhibiting FaMADS1 expression by the active binding of transcript FaTRAB1 with the FaMADS1 promoter. [ABSTRACT FROM AUTHOR]

In this paper, a mechanical model of the deflection dual-stator switched reluctance generator (DDSRG) is developed, and the advantages of the dual-stator structure for the deflecting motion are analyzed. Secondly, the spatio-temporal and spatial distribution characteristics of the inhomogeneous electromagnetic force are derived analytically and further verified by fast Fourier transform (FFT).Thirdly, the spatial and temporal distributions of electromagnetic forces of DDSRG are calculated based on finite element software, and the distributions of electromagnetic forces under different motion states are analyzed. By combining the analysis of modal analysis and harmonic response analysis, the free mode and vibration response acceleration variation laws of the internal and external stator are determined. The results show that the order of electromagnetic forces on the stator at rated speed is mainly 8 times the fundamental frequency, and the modal vibration order is more violent in the order of 2–7. Finally, the experimental platform of DDSRG is built, and the vibration characteristics are tested to verify the validity and accuracy of the proposed simulation results. [ABSTRACT FROM AUTHOR]

Ultra-thin wearing course (UTWC) as an asphalt overlay is widely used in pavement maintenance for extending pavement service life. Researchers focused on improving and evaluating its performance, yet few researchers compare the performance of typical UTWCs. Moreover, some traditional asphalt mixture tests are improper for UTWC due to the thicknesses of UTWC, which is thinner than the traditional asphalt overlay. This study further evaluated the advantages and disadvantages of typical UTWCs. A series of tests were conducted to compare the comprehensive performance of three typical UWTC products, including SMA-10, Novachip-B, and GT-10. Moreover, this study improved the rutting test to evaluate its rutting performance more accurately. Rutting specimens of 20 mm thick and 50 mm thick composite specimens (20 mm UTWC + 30 mm Portland cement concrete slabs) were prepared. Two types of PCC slabs were used, including unprocessed PCC slabs and PCC slabs with preset cracks. The test results showed that Novachip-B showed the best water stability and weakest raveling resistance, while GT-10 showed the best fatigue and anti-skid performance. The rutting performance of UTWCs was reduced because of the influence of preset cracks. The rutting depth of GT-10 was only 60–90% of that of others, showing the comprehensive performance of GT-10 was better than that of others. These results provide a significant reference for the research and application of UTWC. [ABSTRACT FROM AUTHOR]

Remaining life is an important indicator of pavement residual effective service time and is directly related to maintenance decision-making with limited funds. This paper proposes a fast and non-destructive model to predict the remaining life of rigid PCC (Portland cement concrete) pavement, with or without asphalt overlay. Firstly, a model was constructed according to the current Chinese design specifications for concrete pavement integrating an inverse design concept. Secondly, the prediction model was applied to three typical pavement sections with 1430, 1250 and 1000 slabs, respectively. Ground penetrating radar (GPR) was utilized to determine the geometric parameters in the predictive model and the physical state of the pavement. A falling weight detector (FWD) was utilized for determination of the mechanical parameters. A more reasonable equivalent elastic modulus of foundation was back-calculated instead of using the limited model in the design specification. Thirdly, the remaining life was predicted based on the current mechanical and geometric parameters. The distributions of the remaining life of the three pavement sections was statistically analyzed. Finally, a decision-making system to inform maintenance strategy was proposed based on the remaining life and the technical condition of each slab. The results showed that the relationship between the remaining life and the mechanical parameters, geometric parameters and the physical state of the pavement was highly consistent with engineering experience. The success rate of the prediction model was as high as 96%. The proposed fast and non-destructive prediction model showed good engineering applicability and feasibility. The decision-making system was shown to be feasible in terms of economic benefits. [ABSTRACT FROM AUTHOR]

Deflection switched reluctance motors (DSRM) are prone to chattering at low speeds, which always affects the output efficiency of the DSRM and the mechanical loss of the motor. Combining the characteristics of a traditional reluctance motor with the strong nonlinear and high coupling of the DSRM, a control system for a DSRM based on a flexible neural network (FNN) is proposed in this paper. Based on the better robustness and fault tolerance of fuzzy PI control, the given speed signal is adjusted and converted into a torque control signal. As a result, the FNN control module possesses the strong self-learning ability and adaptive adjustment ability necessary to obtain the control voltage signal. Through simulations and experiments, it was verified that the control system can run stably on DSRM and shows good dynamic performance and anti-interference ability. [ABSTRACT FROM AUTHOR]

In order to meet security requirements of space on orbit service, an obstacle avoidance trajectory planning method using improved particle swarm optimization had been presented in this paper. On the basis of the actual overall structure of 7 degrees of freedom redundant space manipulator and the characteristics of obstacles, the envelope method was used to model the arm and obstacles, respectively. The limit conditions to avoid the collision between them were analyzed. Then, the fitness function under the symmetrical conditions of avoiding the collision and searching for the shortest trajectory was constructed. In addition, the obstacle avoidance trajectory planning was solved based on improved particle swarm optimization (IPSO). Finally, simulation experiments were carried out to prove its effectiveness and rationality, where there were symmetrical advantages in two aspects. It can be concluded that the presented method based on IPSO has strong robustness. [ABSTRACT FROM AUTHOR]

A deeper game-tree search can yield a higher decision quality in a heuristic minimax algorithm. However, exceptions can occur as a result of pathological nodes, which are considered to exist in all game trees and can cause a deeper game-tree search, resulting in worse play. To reduce the impact of pathological nodes on the search quality, we propose an iterative optimal minimax (IOM) algorithm by optimizing the backup rule of the classic minimax algorithm. The main idea is that calculating the state values of the intermediate nodes involves not only the static evaluation function involved but also a search into the future, where the latter is given a higher weight. We experimentally demonstrated that the proposed IOM algorithm improved game-playing performance compared to the existing algorithms. [ABSTRACT FROM AUTHOR]

The high density, large capacity, and long-term stability of DNA molecules make them an emerging storage medium that is especially suitable for the long-term storage of large datasets. The DNA sequences used in storage need to consider relevant constraints to avoid nonspecific hybridization reactions, such as the No-runlength constraint, GC-content, and the Hamming distance. In this work, a new nonlinear control parameter strategy and a random opposition-based learning strategy were used to improve the Harris hawks optimization algorithm (for the improved algorithm NOL-HHO) in order to prevent it from falling into local optima. Experimental testing was performed on 23 widely used benchmark functions, and the proposed algorithm was used to obtain better coding lower bounds for DNA storage. The results show that our algorithm can better maintain a smooth transition between exploration and exploitation and has stronger global exploration capabilities as compared with other algorithms. At the same time, the improvement of the lower bound directly affects the storage capacity and code rate, which promotes the further development of DNA storage technology. [ABSTRACT FROM AUTHOR]

Friction stir lap welding (FSLW) of 6061-T6 aluminium sheet and DX51D galvanized steel sheet was carried out by adding zinc foil to the lap interface and studying the influence of the zinc foil on the formation mechanisms and mechanical properties. The influence of the thickness of zinc foil, the plunge depth of the shoulder and the shape of the tools on the mechanical properties of the weld are discussed. Zinc foil reduced the generation of brittle intermetallic compounds, such as Fe4Al13. During the welding process, the axial force was small due to the high rotating speed. Liquid zinc was retained at the interface, where eutectic Al–Zn with low melting point and an Fe–Zn compound were generated to achieve the metallurgical combination of aluminium and steel. The fracture was located in the heat affected zone (HAZ) of the 6061-T6 base aluminium. The results showed that when the zinc foil was too thin, there was less zinc content at the interface; the resulting Al–Zn eutectic had low melting point, was not fully spread and had poor continuity, resulting in poor mechanical properties. When the zinc foil was too thick, a large amount of zinc-based solid solution was generated at the interface, and most of the fracture occurred in the zinc-rich layer. [ABSTRACT FROM AUTHOR]