Road extraction from remote sensing images is crucial for applications such as autonomous driving and urban planning. However, existing methods often neglect the spatial correlation between roads and adjacent buildings, limiting extraction accuracy. This study proposes SAM BE Road, a road-building co-extraction framework based on the Segment Anything Model (SAM), which optimizes road extraction performance by integrating building spatial information. Extensive experiments are carried out on the Massachusetts dataset and the New York dataset, and the results demonstrate that our SAM BE Road outperforms other state-of-the-art methods in extraction accuracy and topological connectivity. The road labels extracted by our method exhibit preferable connectivity, especially in complex urban environments.

Time Series Classification (TSC) has achieved significant results in improving the efficiency and safety of life and production. This paper aims to enhance the accuracy of TSC by encoding one-dimensional time series into two-dimensional Recurrence Plots (RP) to obtain richer texture information. However, existing RPs face issues such as multi-scale issues, tendency confusion, and information redundancy, which hinder their application in TSC tasks. To address these problems, this paper proposes a Large Receptive Field Network (LRFN). LRFN advocates for extracting global information by enlarging the receptive field to overcome the multi-scale issues of RPs. It further addresses other problems by applying Multi-scale Signed RP (MSRP). The method encodes the time series as MSRP and then constructs multiple Global-Local Convolution (GLC) modules. These modules obtain medium, local, and global receptive fields through the Dilated Residual Module, SelfCalibrated Convolution, and Atrous Spatial Pyramid Pooling, respectively. By integrating three different qualities and sizes of receptive fields, comprehensive global information is acquired to address the multi-scale problem. Experimental results show that LRFN can further improve the accuracy of TSC tasks. LRFN demonstrated the best comprehensive performance in comparison to 13 baseline methods across 43 different domains in the UCR dataset. It performed optimally on 20 datasets, with an average accuracy of 91.3%. Additionally, receptive field visualization experiments further validate the effectiveness of LRFN.

Multivariate time series (MTS) classification is essential in industries, such as healthcare and manufacturing, where it helps extract key features from complex data for decision-making and predictions. However, current MTS classification methods often struggle with high-dimensional data and limited labeled samples, leading to poor performance. To address this, we propose a semisupervised MTS classification method via heterogeneous graph representation. This method provides a new perspective on modeling MTS, which integrates various additional information and captures their relationships. First, we employ a contrast temporal self-attention module to obtain sparse MTS representations. Then, we use soft dynamic time warping to model similarities and construct a similarity graph. Next, we learn shapelets and incorporate both subject features and shapelets into the graph, refining it into a heterogeneous graph. Finally, we apply a dual-level graph attention network for node classification task. Experiments on 19 MTS datasets show that our method outperforms existing state-of-the-art methods, demonstrating its effectiveness for MTS classification task.

Multivariate time series classification (MTSC) is a significant research topic in the realm of data mining, with broad applications in different industries, including healthcare, finance, meteorology, and traffic. While existing studies have designed many classifiers based on LSTMs, CNNs, and Transformer, the sophisticated architectures raise concerns regarding efficiency in computation. Additionally, most methods concentrate on a single dimension, typically temporal patterns, without fully considering multi-dimensional information such as the independence and interactions across variables that are essential in multivariate settings. To address these challenges, this article introduces FreConvNet, a lightweight convolutional network integrated with frequency adaptive learning. Inheriting the modular design paradigm of Transformer to achieve multi-view modeling of multivariate time series. FreConvNet consists of two key components: the frequency adaptive block (FAB) and the convolutional feed-forward network (ConvFFN). The FAB leverages the Fourier Transform in conjunction with adaptive filters to capture both long-term and short-term dependencies in the temporal dimension. Following that, ConvFFN captures cross-variable and cross-feature interactions by controlling inter-channel information flow through grouped pointwise convolutions, while introducing non-linearity to enhance representational capacity. Extensive experiments conducted on the well-known UEA archive validate that FreConvNet outperforms existing convolution-based, Transformer-based, and hybrid methods in classification performance and offers a computationally efficient solution.

Recently, time series classification has attracted significant interest. One of the most promising recent approaches is the shapelet transform, which offers two main advantages over traditional approaches: optimization of the shapelet selection process and the flexible integration of different classifiers. However, the high time complexity of identifying shapelets hinders its application in real-time data processing. To overcome this drawback, we propose an adaptive shapelet selection algorithm (ASS). In our method, we first identify Import Data Points (IDPs) for every time series and select the subsequences between two different IDPs as shapelet candidates. We then adaptively select the k best shapelets using ASS. Our experimental results demonstrate that ASS outperforms all other relevant classification methods.

With the development of human-computer interaction and mobile sensors, emotion recognition based on physiological signals has aroused a lively discussion among scholars. The main difficulty faced is the small amount of data which leads to poor training results. In this paper, we proposed a multimodal emotion recognition using CNN-SVM and data augmentation (CSDAMER). Electrocardiography (ECG), galvanic skin response (GSR) and respiration (RSP) are utilized as input data, which are less requiring on the collection environment and can be collected by mobile sensors. To improve the training effect of model, data augmentation is performed by transformations, such as inversion, recombination and noise injection. Moreover, the convolutional layer of the convolutional neural network (CNN) is leveraged to extract the high-level features of the physiological signals, and then the features are input into the support vector machine (SVM) classifier to obtain the recognition results. The experimental results show that CSDAMER achieves 80.7% and 79.92% accuracy in arousal and valance, respectively. Compared with CNN alone, the accuracy of arousal and valance is increased by 12.87% and 9.95%. Meanwhile, the addition of the data augmentation improves the accuracy in arousal and valance by 21.94% and 25.73%.

Emotion recognition has become a research focus in the brain–computer interface and cognitive neuroscience. Electroencephalogram (EEG) is employed for its advantages as accurate, objective, and noninvasive nature. However, many existing research only focus on extracting the time and frequency domain features of the EEG signals while failing to utilize the dynamic temporal changes and the positional relationships between different electrode channels. To fill this gap, we develop the dynamic differential entropy and brain connectivity features based EEG emotion recognition using linear graph convolutional network named DDELGCN. First, the dynamic differential entropy feature which represents the frequency domain feature as well as time domain feature is extracted based on the traditional differential entropy feature. Second, brain connectivity matrices are constructed by calculating the Pearson correlation coefficient, phase-locked value and transfer entropy, and then are used to denote the connectivity features of all electrode combinations. Finally, a linear graph convolutional network is customized and applied to aggregate the features from total electrode combinations and then classifies the emotional states, which consists of five layers, namely, an input layer, two linear graph convolutional layers, a fully connected layer, and a softmax layer. Extensive experiments show that the accuracies in the valence and arousal dimensions reach 90.88% and 91.13%, and the precision reaches 96.66% and 97.02% on the DEAP dataset, respectively. On the SEED dataset, the accuracy and precision reach 91.56% and 97.38%, respectively.

Although more and more researchers pay attention to the emotion classification, traditional emotion classification methods can not embrace changes in the global and local areas of the human brain after being stimulated. We propose an emotion classification method based on SVM combining brain functional connectivity. Firstly, the nonlinear phase-locked value (PLV) is used to calculate the multiband brain functional connectivity network, which is then converted into a binary brain network, and seven features of binary brain network are calculated. Secondly, support vector machines (SVM) are used to classify positive and negative emotions at the valence dimension and arousal dimension in the multiband. Experimental results on DEAP show that the best emotion classification accuracy of the proposed method is 86.67% in the arousal dimension, and 84.44% in the valence dimension. The results demonstrate that the classification accuracy of the arousal dimension is better than the valence dimension and the Beta2 frequency band is more suitable for emotion classification. Finally, several findings on brain functional connectivity network is discussed. The left and right areas of brain functional connectivity network are unbalanced in the low frequency band, and the feature values of clustering coefficient, average shortest path length, global efficiency, local efficiency, node degree are positively correlated with the arousal degree in the arousal dimension. Humans emotions are suppressed in the low frequency band, and the brain functional connectivity network after emotional stimulation is strengthened in the high frequency band. Our findings on emotion classification are valuable and consistent with the study of neural mechanisms.

Summary The application of anomaly detection to data monitoring is a fundamental requirement of the public service systems of a smart city. Many detection methods have been proposed for identifying anomalous situations, including methods based on periodicity or biseries correlations. However, the detection results of these methods are not ideal. Thus, we present a new anomaly detection algorithm for time series based on the relative outlier distance (ROD) and biseries correlations. The proposed algorithm detects outliers based on the ROD and identifies abnormal points and change points based on biseries correlations. Experimental results show that our method achieves better recall and F1‐measure scores than various time series–based techniques while maintaining a high level of precision.

Electronic health records are digital records of patients' medical history, diagnosis, medication, treatment plans. EHRs not only contain the patients' medical and treatment history, but also systematically collect patients' clinical data. Therefore, it is very valuable to improve the patient's health care management by mining the information in the EHRs. However, due to the irregularities and sparsity of EHRs, EHRs mining is very challenging. In this paper, the laboratory data, physiological indicators and diagnosis time during the patients' hospitalization period are extracted from the MIMIC-III database. Then, the extracted features are used to generate the patients' representation vector. Finally, we propose a prediction model based on BiLSTM and attention mechanism, which is called Bi-Attention. The BiLSTM is adopted to learn the forward and backward timing information in the patient's representation vectors and to predict the patient's disease by utilizing the specific clinical information in the timed medical record with the attention mechanism. The experimental results show that compared with other methods, the proposed model can effectively improve the prediction performance.

A serious challenge that confronts shapelet-based algorithms for time series classification is finding optimal shapelets in a short time. Representative shapelet-discovery algorithms find shapelets by evaluating the qualities of candidates extracted from the subsequences. One of the main difficulties is the large amount of time consumed, due to the excessive number of shapelet candidates. To address the above problem, in this paper we propose a fast and interpretable candidate-extraction algorithm to accelerate the process of shapelet discovery. The proposed algorithm utilizes a time series subclass splitting technique to sample time series dataset first. Then, an IDP (Important Data Point)-based selective-extraction strategy is used to extract shapelet candidates. The generated candidates have significant improvements in quality and reductions in quantity. Furthermore, the shapelet candidates generated are more interpretable. To test the effectiveness of the shapelet candidates generated, we transform the original time series and use an off-the-shelf attribute-selection technique to select optimal shapelets from candidates. We then evaluate the proposed algorithm through extensive experiments. The results demonstrate that the proposed algorithm makes significant improvements in accuracy, compared with baselines. Meanwhile, the time consumption is also greatly reduced.

In the case of a cloud-based remote control system such as SCADA (Supervisory Control and Data Acquisition) that enables users to collect data from cloud-connected machines deployed anywhere at any time. However, machine data models may not be updated in a timely manner after the devices are upgrades or modified. This leads to mismatches between the machine data and data models. A key obstacle of the matching is that the machines can be modified. To address this, we present a self-evolving method for machine data model. We give the description of the evolution of machine data models and the self-evolving method for the models in details. The method detects the conflicts between the machine data and models, and transfer or derive models if necessary. Our method can thus facilitate the evolution of machine data models and ensure that every machine in the cloud corresponds to the correct machine data model automatically. At last, we present two case studies to validate our method.

With the development of intelligent manufacturing technology, it can be foreseen that time series data generated by smart devices will raise to an unprecedented level. For time series with high amount, high dimension and renewal speed characteristics, resulting in difficult data mining and presentation on the original time series data. This paper presented a piecewise linear representation based on importance data points for time series data, which called PLR_IDP for short. The method finds importance data points by calculating the fitting error of single point and piecewise, and then represents time series approximately by linear composed of the importance data points. Results from theoretical analysis and experiments show that PLR_IDP reduces the dimensionality, holds the main characteristic with small fitting error of segments and single points.

The kNN join problem, denoted by R ×KNNS, is to find the k nearest neighbors from a given dataset S for each point in the query set R. It is an operation required by many big data applications. As large volume of data are continuously generated in more and more real-life cases, we address the problem of monitoring kNN join results on data streams. Specifically, we are concerned with answering kNN join periodically at each snapshot which is called snapshot kNN join. Existing kNN join solutions mainly solve the problem on static datasets, or on a single centralized machine, which are difficult to scale to large data on data streams. In this paper, we propose to incrementally calculate the kNN join results of time tifrom the results of snapshot ti-1. Typically, for the data continuously generated on the data stream, we can get Si= Si-1+ ΔSifor the valid datasets of adjacent snapshots, where ΔSidenotes the updated points between time ti-1and ti. Our basic idea is to first find the queries in R whose kNN results can be affected by the updated points in ΔSi, and then update the kNN results of these small part of queries respectively. In this way, we can avoid calculating the kNN join results on the whole dataset Siin time ti. We propose an implementation of searching for affected query points in MapReduce to scale to large volume of data. In brief, the mappers partition the datasets into groups, and the reducers search for affected queries separately on each group of points. Furthermore, we present the enhanced strategies of data partitioning and grouping to reduce the shuffling cost and computational cost. Extensive experiments on real-world datasets demonstrate that proposed methods are efficient, robust, and scalable.