Search

Correlated time series (CTS) forecasting is essential in many practical applications, such as traffic management and server load control. Various deep learning based solutions have been proposed to improve forecasting accuracy. However, while models have become increasingly computationally intensive, they struggle to improve accuracy. This study aims instead to enable more lightweight, accurate models suitable for resource-constrained devices. To achieve this goal, we characterize popular CTS forecasting models, yielding two observations for developing lightweight CTS forecasting. On this basis, we propose the LightCTS framework that adopts plain stacking of temporal and spatial operators instead of alternate stacking which is much more computationally expensive. Moreover, LightCTS features light temporal and spatial operators, L-TCN and GL-Former, offering improved computational efficiency without compromising their feature extraction capabilities. LightCTS also encompasses a last-shot compression scheme to reduce redundant temporal features and speed up subsequent computations. Next, we equip LightCTS with two knowledge distillation modules, Tafd and Caad, that result in LightCTS$^\star$★ retaining the original benefits of LightCTS, while also being able to adapt to varying levels of ultra-constrained resources. Experimental studies offer detailed insight into these proposals and provide evidence that both LightCTS and LightCTS$^\star$★ are capable of nearly state-of-the-art accuracy at substantially reduced computational costs.

Train delay prediction is a key technology for intelligent train scheduling and passenger services. We propose a train delay prediction model that takes into account the asynchrony of train events, the dynamics of train operations, and the diversity of influencing factors. Firstly, we consider train operations as discrete sequences of train events and propose a train arrival neural temporal point process (TANTPP) framework focused on predicting train delays that explicitly models the asynchrony of train events. Secondly, we introduce a multi-source dynamic spatiotemporal embedding method for the feature encoder in the TANTPP framework, which enhances the capability to capture the features of train operation networks. Thirdly, to better capture the distribution of train events in the TANTPP framework, we utilize a log-normal mixture hybrid method to learn the probability density distribution of train arrival events. Finally, the experimental result on real-world datasets demonstrates that the TANTPP model outperforms current state-of-the-art models, reducing the MAE by 10.85%, the RMSE by 9.8%, the RRSE by 3.78% and the MAPE by 10.11% on average. To the best of our knowledge, this is the first study to utilize neural temporal point processes to enhance train delay prediction.

Uncertainty is an essential consideration for time series forecasting tasks. In this work, we focus on quantifying the uncertainty of traffic forecasting from a unified perspective. We develop a novel traffic forecasting framework, namely Deep Spatio-Temporal Uncertainty Quantification (DeepSTUQ), which can estimate both aleatoric and epistemic uncertainty. Specifically, we first leverage a spatio-temporal model to model the complex spatio-temporal correlations of traffic data. Subsequently, two independent sub-neural networks maximizing the heterogeneous log-likelihood are developed to estimate aleatoric uncertainty. To estimate epistemic uncertainty, we combine the merits of variational inference and deep ensembling by integrating the Monte Carlo dropout and the Adaptive Weight Averaging re-training methods, respectively. Furthermore, to relax the Gaussianity assumption, mitigate overfitting, and improve horizon-wise uncertainty quantification performance, we define a new calibration method called Multi-horizon Conformal Calibration (MHCC). Finally, we provide a theoretical analysis of the proposed unified approach based on the PAC-Bayes theory. Extensive experiments are conducted on four public datasets, and the empirical results suggest that the proposed method outperforms state-of-the-art methods in terms of both point prediction and uncertainty quantification.

Multi-source cross-project defect prediction (MSCPDP) attempts to transfer defect knowledge learned from multiple source projects to the target project. MSCPDP has drawn increasing attention from academic and industry communities owing to its advantages compared with single-source cross-project defect prediction (SSCPDP). However, two main problems, which are how to effectively extract the transferable knowledge from each source dataset and how to measure the amount of knowledge transferred from each source dataset to the target dataset, seriously restrict the performance of existing MSCPDP models. In this paper, we propose a novel multi-source transfer weighted ensemble learning (MASTER) method for MSCPDP. MASTER measures the weight of each source dataset based on feature importance and distribution difference and then extracts the transferable knowledge based on the proposed feature-weighted transfer learning algorithm. Experiments are performed on 30 software projects. We compare MASTER with the latest state-of-the-art MSCPDP methods with statistical test in terms of famous effort-unaware measures (i.e., PD, PF, AUC, and MCC) and two widely used effort-aware measures ($P_{opt}20\%$Popt20% and IFA). The experiment results show that: 1) MASTER can substantially improve the prediction performance compared with the baselines, e.g., an improvement of at least 49.1% in MCC, 48.1% in IFA; 2) MASTER significantly outperforms each baseline on most datasets in terms of AUC, MCC, $P_{opt}20\%$Popt20% and IFA; 3) MSCPDP model significantly performs better than the mean case of SSCPDP model on most datasets and even outperforms the best case of SSCPDP on some datasets. It can be concluded that 1) it is very necessary to conduct MSCPDP, and 2) the proposed MASTER is a more promising alternative for MSCPDP.

Emotion, a fundamental trait of human beings, plays a pivotal role in shaping aspects of our lives, including our cognitive and perceptual abilities. Hence, emotion recognition also is central to human communication, decision-making, learning, and other activities. Emotion recognition from electroencephalography (EEG) signals has garnered substantial attention due to advantages such as noninvasiveness, high speed, and high temporal resolution; driven also by the complementarity between EEG and other physiological signals at revealing emotions, recent years have seen a surge in proposals for EEG-based multimodal emotion recognition (EMER). In short, EEG-based emotion recognition is a promising technology in medical measurements and health monitoring. While reviews exist, which explore emotion recognition from multimodal physiological signals, they focus mostly on general combinations of modalities and do not emphasize studies that center on EEG as the fundamental modality. Furthermore, existing reviews take a methodology-agnostic perspective, primarily concentrating on the biomedical basis or experimental paradigms, thereby giving little attention to the methodological characteristics unique to this field. To address these gaps, we present a comprehensive review of current EMER studies, with a focus on multimodal machine learning models. The review is structured around three key aspects: multimodal feature representation learning, multimodal physiological signal fusion, and incomplete multimodal learning models. In doing so, the review sheds light on the advances and challenges in the field of EMER, thus offering researchers who are new to the field a holistic understanding. The review also aims to provide valuable insight that may guide new research in this exciting and rapidly evolving field.

With the diffusion of online mobile devices with geo-location capabilities, the infrastructure necessary for real-world deployment of Spatial Crowdsourcing (SC), where so-called mobile workers are assigned location-sensitive tasks, is in place. Some SC tasks cannot be completed by a single worker due to their complexity, but rather must be assigned to and completed by a group of users. Achieving such group assignments that satisfy all group members evenly is an open challenge. To address this challenge, we propose a novel preference-aware group task assignment framework encompassing two components: Mutual Information-based Preference Modeling (MIPM) and Preference-aware Group Task Assignment (PGTA). The MIPM component learns the preferences of groups contrastively by maximizing the mutual information between workers and worker groups based on worker-task and group-task interaction data and by using an attention mechanism to weight group members adaptively. In addition, curriculum negative sampling is adopted to generate a small number of negative workers for each worker group, following the principles of curriculum learning. Next, the PGTA component offers an optimal task assignment algorithm that employs tree decomposition to assign tasks to appropriate worker groups, with the aim of maximizing the number of task assignments while prioritizing more interested groups when assigning tasks. The task assignment framework also features preference-constrained pruning of unpromising worker groups to speed up the assignment process. Finally, we report extensive experiments that offer evidence of the effectiveness and practicality of the paper's proposal.

Statins are considered the cornerstone of secondary prevention in patients with atherosclerotic cardiovascular disease (ASCVD). However, many patients fail to achieve the guide-recommended goal of low-density lipoprotein cholesterol (LDL-C) after statin monotherapy, leading to a high residual risk of cardiovascular events. Owing to individual differences in statin therapy, it is possible first to consider changing the type of statin before adding nonstatin medications in certain patients to improve LDL-C management. We developed and evaluated a statin recommendation system using real-world data. Ensemble learning was performed to develop the recommendation system that integrated the output results of support vector machines (SVM) and the similarity of patients. Model performance was assessed to investigate whether treatment according to the recommended model would increase the proportion of patients with the primary end point. Finally, a total of 3510 patients were enrolled in the development and validation of the recommender system. Of them, 1240 patients received atorvastatin (35.3%), 1714 patients received rosuvastatin (48.8%), and 556 patients received pitavastatin (15.8%). The statin recommendation system could significantly improve LDL-C target rate achievement in the recommended treatment group compared with the nonrecommended treatment group in the validation set (50.8% vs. 31.5%, P< 0.001). This study demonstrated that the statin recommendation system could significantly improve the achievement of LDL-C goals in ASCVD patients, providing a new approach to improve LDL-C management.

Thermal stratification in Lead-based cooled Fast Reactors (LFRs) may cause thermal stress and thermal fatigue of the reactor components, thereby jeopardizing the operating safety. This study presents a thermal stratification analysis of the primary system of LFRs, utilizing the computational fluid dynamics method to assess both the steady-state condition and the loss of flow accident. Under the steady-state condition, thermal stratification is observed within the upper plenum and upper region of the cold pool of the LFRs, but it is not pronounced enough to pose a a significant threat to the structural safety of the reactor. Under the loss of flow accident, the thermal stratification phenomenon in the upper plenum and the cold pool exhibits a pattern of rapid intensification followed by gradual relaxation over time. Specifically, stratification interface in upper plenum persists and stratification number increases, while in the cold pool stratification interface decreases and stratification number remains constant. Furthermore, stratification number in the upper plenum is more than that in the cold pool, indicating that thermal stratification is more pronounced in the upper plenum. This study serves as a valuable reference the analyzing thermal stratification in the primary system of LFRs.

At present, the manufacturer collaborates on selling and recycling through an online platform. We explores the mode selection of closed-loop supply chain (CLSC) under the reselling or agency modes in forward SC and under the delegated-recycling mode or agency-recycling mode in reverse SC considering data intelligence level with an online platform. The results show that both the retail price of intelligent connected products (ICPs) and the demand of new ICPs are positively correlated with data intelligence level of new ICPs. The recycling price of old ICPs is negatively relevant to data intelligence level of old ICPs. However, the recycling quantity of old ICPs is positively related to data intelligence level of old ICPs. Increasing data intelligence level of new ICPs is benefit to enterprises. Moreover, it is very unconventional that the demand of new ICPs is increasing when the retail price of new ICPs increases. In addition, the manufacturer can choose the optimal selling mode and recycling mode by setting the selling commission rate and the recycling commission rate.

Graphical Abstract:

High-speed train delay prediction has always been one of the important research issues in the railway dispatching. Accurate and interpretable delay prediction can enable staff to implement preventive measures and scheduling decisions in advance, and guide relevant departments to cooperate in completing complex transportation tasks, so as to improve rail transit operations, service quality, and the efficiency of train operation. This article proposes a new interpretable model based on graph community neural network and time-series fuzzy decision tree. This model can well capture the influence of spatiotemporal characteristics, train community structure, and multifactor in high-speed train station delay prediction. Besides, the time series fuzzy decision tree based on multiobjective optimization and reduced error pruning can mine potential decision rules to improve the model's interpretability, transparency, and high reliability. Finally, we prove that the prediction effect of the proposed model is superior than the other seven state-of-the-art models and our model is interpretable.

Aircraft are developing towards the goal of low energy consumption and high efficiency; higher requirements are put forward for the energy utilization efficiency of the environmental control system. This paper presents a new type of closed air circulation system driven by an electric compressor, which can make more efficient use of aircraft energy. Aiming at the new environmental control system, the neural network algorithm was used to construct the system component model and then carry out the system simulation. The results showed that the average error between the simulation outputs and experimental values of the model constructed by this method was less than 4.5%, which effectively verifies the correctness of the system modeling. Compared with the traditional three-wheel system, the proposed environmental control system driven by an electric compressor can reduce the engine air intake by 64.5% and improve the cooling efficiency by 210%. The proposed new environmental control system can provide a reference for the design of the environmental control system of the next generation of aircraft.