Sliding Time Window Research Articles

PREDIKSI INFLOW DAN OUTFLOW UANG KARTAL DI PROVINSI BALI DENGAN METODE NEURO-FUZZY

In this paper, we present a novel approach to data-driven neuro-fuzzy modeling, which aims to create accurate monthly inflow and outflow forecast of money (M0) in Bali Province. The data is monthly time series included some religious ceremony identification variables and a monthly dummy variable from January 2011 to March 2019. Well known, Bali Province has unique cultures, the only one province which Hinduism majority religion in Indonesia, and listed as top tourism destination in the world. The neuro-fuzzy models were created using ANFIS architecture and sliding window time series analysis, then simulated using walk forward validation, interpreted using MAPE, and NRMSE. Based on the simulation of the last 24 months, the model of inflow obtained MAPE 23.33% (worth considering) and NRMSE 18.68% (accurate). Meanwhile, the model of outflow obtained MAPE 19.24% (accurate) and NRMSE 8.71% (very accurate). These models and their pieces of information could assist the central bank in Bali Province to prepare cash for money (M0) outflow and managed technic for counting money (M0) inflow.

Dynamic Group Recommendation Algorithm Based on Member Activity Level

The rapid development of social networks has led to an increased desire for group entertainment consumption, making the study of group recommender systems a hotspot. Existing group recommender systems focus too much on member preferences and ignore the impact of member activity level on recommendation results. To this end, a dynamic group recommendation algorithm based on the activity level of members is proposed. Firstly, the algorithm predicts the unknown preferences of members using a time-series-oriented rating prediction model. Secondly, considering the dynamic change of member activity level, the group profile is generated by designing a sliding time window to investigate the recent activity level of each member in the group at the recommended moment, and preference is aggregated based on the recent activity level of members. Finally, the group recommendations are generated based on the group profile. The experimental results show that the algorithm in this paper achieves a better recommendation result.

Study on the heterogeneity in mixture carrageenan gels viewed by long time particle tracking

The network structure of mixed κ- (KC) and ι- (IC) carrageenan gels were investigated using particle tracking in a long time observation. The Brownian motion of 100 nm particles tracked for ~1400 s was utilized to probe the microenvironment in the mixed gels. The individual mean square displacement (msd) of particles in different mixing ratios of KC/IC showed a broad distribution, indicating spatial heterogeneity in the particle diffusion, suggesting the phase-separated network structure with KC-rich and IC-rich domains. The broad distribution of msd further suggests fast-moving particles in IC-rich and slow-moving particles in KC-rich domains. The distribution of displacement for each of the fast-moving particles showed a non-Gaussian distribution indicating the heterogeneous diffusion behavior. The time-course of the radius of gyration in a sliding time window with a duration time for the mobile particles revealed multiple confinement events. These were attributed to the sterical hindrance or the interaction with the surface of the KC-rich domain, which tolerates the breakaway from the confinement by the Brownian motion. Simulation of particle diffusion was performed to understand the interaction that causes the confinement period. Results revealed that the confinement was caused by the interaction between the particle and the KC-rich domain covered with KC chains. The KC chains have the negatively charged sulfonic group, but salt ions of KCl should satisfyingly shield the repulsive interaction by the sulfonic group to have cohesive interaction with the particle.

Nonparametric modeling of ship maneuvering motion based on Gaussian process regression optimized by genetic algorithm

A novel method, Gaussian process regression optimized by genetic algorithm (GA-GPR), is proposed for nonparametric modeling of ship maneuvering motion. A genetic algorithm with adaptive crossover and mutation operations is introduced for adjustment and optimization of hyperparameters in the kernel function. The sliding time window method is adopted to update the training samples to improve the adaptability of the model. Taking the Mariner vessel and the container ship KCS as study objects, the simulation data and the real measured data provided by SIMMAN 2020 workshop are employed to evaluate the proposed method. The prediction results of the proposed method and the traditional Gaussian process regression are compared and validated against available experimental data of zigzag and turning circle maneuvers. It shows that the proposed method has higher prediction accuracy and better generalization ability, indicating that the hyperparameters optimized by genetic algorithm make the model closer to the ship's dynamic characteristics.

Robust K-means algorithm with weighted window for seismic facies analysis

Abstract Seismic facies analysis can generate a map to describe the spatial distribution characteristics of reservoirs, and therefore plays a critical role in seismic interpretation. To analyse the characteristics of the horizon of interest, it is usually necessary to extract seismic waveforms along the target horizon using a selected time window. The inaccuracy of horizon interpretation often produces some inconsistent phases and leads to inaccurate classification. Therefore, the developed adaptive phase K-means algorithm proposed a sliding time window to extract seismic waveforms. However, setting the maximum offset of the sliding window is difficult in a real data application. A value that is too large may cause the cross-layer problem, whereas a value that is too small reduces the flexibility of the algorithm. To address this disadvantage, this paper proposes a robust K-means (R-K-means) algorithm with a Gaussian-weighted sliding window for seismic waveform classification. The used weights punish those windows distant from the interpretation horizon in the objective function, consequently producing a smaller range of horizon adjustments even when using relatively large maximum offsets and benefitting the generation of stable and reliable seismic facies maps. The application of real seismic data from the F3 block proves the effectiveness of the proposed algorithm.

Spatial Connectivity and Temporal Dynamic Functional Network Connectivity of Musical Emotions Evoked by Dynamically Changing Tempo.

Music tempo is closely connected to listeners’ musical emotion and multifunctional neural activities. Music with increasing tempo evokes higher emotional responses and music with decreasing tempo enhances relaxation. However, the neural substrate of emotion evoked by dynamically changing tempo is still unclear. To investigate the spatial connectivity and temporal dynamic functional network connectivity (dFNC) of musical emotion evoked by dynamically changing tempo, we collected dynamic emotional ratings and conducted group independent component analysis (ICA), sliding time window correlations, and k-means clustering to assess the FNC of emotion evoked by music with decreasing tempo (180–65 bpm) and increasing tempo (60–180 bpm). Music with decreasing tempo (with more stable dynamic valences) evoked higher valence than increasing tempo both with stronger independent components (ICs) in the default mode network (DMN) and sensorimotor network (SMN). The dFNC analysis showed that with time-decreasing FNC across the whole brain, emotion evoked by decreasing music was associated with strong spatial connectivity within the DMN and SMN. Meanwhile, it was associated with strong FNC between the DMN–frontoparietal network (FPN) and DMN–cingulate-opercular network (CON). The paired t-test showed that music with a decreasing tempo evokes stronger activation of ICs within DMN and SMN than that with an increasing tempo, which indicated that faster music is more likely to enhance listeners’ emotions with multifunctional brain activities even when the tempo is slowing down. With increasing FNC across the whole brain, music with an increasing tempo was associated with strong connectivity within FPN; time-decreasing connectivity was found within CON, SMN, VIS, and between CON and SMN, which explained its unstable valence during the dynamic valence rating. Overall, the FNC can help uncover the spatial and temporal neural substrates of musical emotions evoked by dynamically changing tempi.

Effects of Individual Differences on Measurements’ Drowsiness-Detection Performance

Individual differences (IDs) may reduce the detection-accuracy of drowsiness-driving by influencing measurements’ drowsiness-detection performance (MDDP). The purpose of this paper is to propose a model that can quantify the effects of IDs on MDDP and find measurements with less impact by IDs to build drowsiness-detection models. Through field experiments, drivers’ naturalistic driving data and subjective-drowsiness levels were collected, and drowsiness-related measurements were calculated using the double-layer sliding time window. In the model, MDDP was represented by |Z-statistics| of the Wilcoxon-test. First, the individual driver’s measurements were analysed by Wilcoxon-test. Next, drivers were combined in pairs, measurements of paired-driver combinations were analysed by Wilcoxon-test, and measurement’s IDs of paired-driver combinations were calculated. Finally, linear regression was used to fit the measurements’ IDs and changes of MDDP that equalled the individual driver’s |Z-statistics| minus the paired-driver combination’s |Z-statistics|, and the slope’s absolute value (|k|) indicated the effects of ID on the MDDP. As a result, |k| of the mean of the percentage of eyelid closure (MPECL) is the lowest (4.95), which illustrates MPECL is the least affected by IDs. The results contribute to the measurement selection of drowsiness-detection models considering IDs.

Improving numerical forecast of the rainstorms induced by Mongolia cold vortex in North China with the frequency matching method

Mongolia cold vortex is an important synoptic system causing summer convective precipitation in North China and the numerical model shows poor performances for forecasting the convective rainstorms. It is, therefore, urgent to improve the forecast skill of the rainstorms induced by Mongolia cold vortex. Taking the Haihe River Basin as an example and using the observational precipitation data monitored by automatic meteorological stations and the forecast precipitation data from European Centre for Medium-Range Weather Forecasts (ECMWF), this study is the first time to recombine the numerical forecast series that have the greatest similarity to the observed precipitation by using correlation analysis on a sliding time window. We then corrected the recombinant forecast series by using frequency matching method (FMM) and finally established a correction model for improving precipitation forecast accuracy. Results revealed that FMM could improve the precipitation forecast skill for 1-3 d lead-time. Compared with light, moderate, and heavy rain, the improvement effects are more obvious for rainstorm (50–99.9 mm) and heavy rainstorm (≥100 mm). Moreover, the calibrated forecasts have good performances to improve the accuracy for most magnitudes but not spatial distributions. These findings demonstrate that FMM has good performance for correcting precipitation forecasts induced by Mongolia Cold Vortex and thus can be used to improve the forecast skill of summer rainstorms, providing more accurate forcing inputs for hydrologic model to improve flooding forecast skill in North China.

An online identification approach for a nonlinear ship motion model based on a receding horizon

The ship motion system is a nonlinear control object, and its system parameters exhibit time-varying characteristics with the ship motion state, which increases the difficulty of control. Therefore, parameter identification has an important significance for the stability of ship motion control. Aiming at the real-time identification problem of the nonlinear and time-varying ship motion system during movement, this paper reconstructs the ship motion system with the propeller speed and rudder angle as control variables and designs an online identification algorithm with the sliding time window method based on the extended Kalman filter algorithm. In addition, to solve the problem of noise in ship motion data collected in real-time, a real-time wavelet filter is developed to perform online preprocessing of the input data of the identification algorithm. The applicability of the method is further demonstrated via a model-scale Korea Research Institute of Ships and Ocean Engineering container ship free-running experiments in a basin.

Integration and Segregation of Dynamic Functional Connectivity States for Mild Cognitive Impairment Revealed by Graph Theory Indicators.

Mild cognitive impairment (MCI) is an intermediate stage between normal aging and dementia. Researchers tend to discuss its early state (early MCI, eMCI) due to its high conversion rate of dementia and poor treatment effect in the middle and late stages. Currently, the research on the disease evolution of the brain functional networks of patients with MCI has gradually become a research hotspot. In this study, we compare the differences in dynamic functional connectivity among eMCI, late MCI (lMCI), and normal control (NC) groups, and their graph theory indicators reveal the integration and segregation of functional connectivity states. Firstly, dynamic functional network windows were constructed based on the sliding time window method, and then these window samples were clustered by k-means to extract the functional connectivity states. The differences in the three groups were compared by analyzing the graph theory indicators, such as the participation coefficient, module degree distribution, clustering coefficient, global efficiency, and local efficiency, which distinguish the functional connectivity states. The results reveal that the NC group has the strongest integration and segregation, followed by the eMCI group, and the lMCI group has the weakest integration and segregation. We conclude that with the aggravation of MCI, the integration and segregation of dynamic functional connectivity states tend to decline. The results also reflect that the lMCI group has significantly more brain functional connections in some states, such as IPL.L-MTG.R and DCG.R-SMG.L, than the eMCI group, while the lMCI group has significantly less OLF.L-SPG.L than the NC group.

Intermittent fault detection for linear discrete-time stochastic multi-agent systems

In this paper, the intermittent fault (IF) detection problem is investigated for a class of linear discrete-time stochastic multi-agent systems (MASs). By using the reduced-order observer method and introducing a sliding-time window, the truncated residuals are designed to detect the appearing time and disappearing time of the IF. Moreover, two sets of hypothesis tests are proposed to determine the detection and location thresholds of the IF, and the detectability of the IF is discussed in the framework of statistical analysis. Finally, a simulation example is provided to validate the feasibility and effectiveness of the proposed method.

Optimizing steady-state responses to index statistical learning: Response to Benjamin and colleagues

Neural entrainment refers to the tendency of neural activity to align with an ongoing rhythmic stimulus. Measures of neural entrainment have been increasingly leveraged as a tool to understand how the brain tracks different types of regularities in sensory input. However, the methods used to quantify neural entrainment are varied, with numerous analytic decision points whose consequences have not been well-characterized. In a valuable contribution to this field, Benjamin, Dehaene-Lambertz and Flo (submitted) systematically compare various methodological approaches for studying neural entrainment. They demonstrate that the use of overlapping epochs, in which sliding time windows are extracted and analyzed, results in an artifactual inflation of entrainment estimates at the frequency of overlap. Here, in response to this updated best practice recommendation, we reanalyzed three previously published datasets that had been previously analyzed with overlapping epochs. Although our main results and conclusions are unaltered from those originally reported, we agree with Benjamin and colleagues that overlapping epochs should generally be avoided in classic analyses of steady-state experiments, which aim to quantify overall peaks in phase or power across an entire experimental duration. However, we present a case that overlapping epochs may be beneficial in fine-grained analyses of neural entrainment over time. The use of overlapping epochs in such analyses could improve temporal resolution without complicating interpretability of the results in cases where the question of interest relates to relative changes in neural entrainment over time within a given frequency.

Intermittent fault detection for delayed stochastic systems over sensor networks

This paper is concerned with the intermittent fault (IF) detection problem for a class of linear discrete-time stochastic systems over sensor networks with constant time delay. By utilizing the lifting method, the distributed decoupled observers are proposed based on the output information of neighbor nodes and the node itself. In order to detect the appearing time and disappearing time of the IF, the truncated residuals are designed by introducing a sliding-time window. Furthermore, the IF detection and location thresholds are determined based on the hypothesis testing technique and the detectability of the IF is analyzed in the framework of stochastic analysis. Finally, a simulation example is presented to illustrate the effectiveness of the derived results.

Classification of patients with AD from healthy controls using entropy-based measures of causality brain networks

BackgroundMachine learning and pattern recognition have been widely used in rs-fMRI data to investigate Alzheimer’s disease (AD). However, many previous methods extracted discriminative features based on functional correlations, which may ignore the asynchronous causality influence of neural activities. New methodWe propose a novel method for AD diagnosis using Sample Entropy to measure the neural complexity of the brain causality network. Granger Causality analysis with a sliding time window was applied on rs-fMRI data of 29 AD patients and 30 cognitive normal (CN) controls to compute the whole brain's causality series. We further grouped these causality series into clusters by agglomerative hierarchical clustering algorithm and computed Sample Entropy of the clusters as the classification features. ResultsWe explored four different classifiers, i.e., XGBoost, SVM cluster, Random Forest, and SVM, based on the above features. An accuracy of 89.83%, with a sensitivity of 90.00% and a specificity of 89.66%, was achieved with the optimal feature subsets using the SVM classifier. Comparison with existing methodsWith the same dataset, the performances of the proposed method were generally higher than those of conventional methods for AD classification based on Pearson's correlation network, dynamic Pearson's correlation network, High-order correlation network, and causality correlation network. ConclusionsOur method demonstrates the measure of Sample Entropy with causality connection as a powerful tool to classify AD patients from CN controls, and provides a deep insight into the neuropathogenesis of AD.

Influence of Sliding Time Window Size Selection Based on Heart Rate Variability Signal Analysis on Intelligent Monitoring of Noxious Stimulation under Anesthesia.

In recent decades, little progress of objective evaluation of pain and noxious stimulation has been achieved under anesthesia. Some researches based on medical signals have failed to provide a general understanding of this problem. This paper presents a feature extraction method for heart rate variability signals, aiming at further improving the evaluation of noxious stimulation. In the process of data processing, the empirical mode decomposition is used to decompose and recombine heart rate variability signals, and the sliding time window approach is used to extract the signal features of noxious stimulation, respectively. The influence of window size on feature extraction is studied by changing the window size. By comparing the results, the feature extraction in the process of data processing is valuable, and the selection of window size has a significant impact. With the increase of selected window sizes, we can get better detection results. But for the best choice of window size, to ensure the accuracy of the results and to make it easy to use, then, we need to get just a suitable window size.

An aggregator‐based resource allocation in the smart grid using an artificial neural network and sliding time window optimization

The success of an efficient and effective aggregator-based residential demand response system in the smart grid relies on the day-ahead customer incentive pricing (CIP) and the load shifting protocols. An artificial neural network model is designed to generate the day-ahead CIP for the aggregator based on historical data. Load scheduling is proposed as a day-ahead optimization problem that is solved using a blocked sliding window technique using parallel computing. With the assumptions made, the proposed algorithm improved the aggregator performance by reducing the overall simulation time from 275 to 45 min and increasing the aggregator forecast profits and customer savings by 11.85% and 35.99% compared to the previous genetic algorithm-based approach.

Dynamic financial distress prediction based on class-imbalanced data batches

This study proposes two approaches for dynamic financial distress prediction (FDP) based on class-imbalanced data batches by considering both concept drift and class imbalance. One is based on sliding time window and synthetic minority over-sampling technique (SMOTE) and the other is based on sliding time window and majority class partition. Support vector machine, multiple discriminant analysis (MDA) and logistic regression are used as base classifiers in the experiments on a real-world dataset. The results indicate that the two approaches perform better than the pure dynamic FDP (DFDP) models without class imbalance processing and the static FDP models either with or without class imbalance processing.

State-independent and state-dependent patterns in the rat default mode network

Resting-state studies have typically assumed constant functional connectivity (FC) between brain regions, and these parameters of interest provide meaningful descriptions of the functional organization of the brain. A number of studies have recently provided evidence pointing to dynamic FC fluctuations in the resting brain, especially in higher-order regions such as the default mode network (DMN). The neural activities underlying dynamic FC remain poorly understood. Here, we recorded electrophysiological signals from DMN regions in freely behaving rats. The dynamic FCs between signals within the DMN were estimated by the phase locking value (PLV) method with sliding time windows across vigilance states [quiet wakefulness (QW) and slow-wave and rapid eye movement sleep (SWS and REMS)]. Factor analysis was then performed to reveal the hidden patterns within the DMN. We identified distinct spatial FC patterns according to the similarities between their temporal dynamics. Interestingly, some of these patterns were vigilance state-dependent, while others were independent across states. The temporal contributions of these patterns fluctuated over time, and their interactive relationships were different across vigilance states. These spatial patterns with dynamic temporal contributions and combinations may offer a flexible framework for efficiently integrating information to support cognition and behavior. These findings provide novel insights into the dynamic functional organization of the rat DMN.

A soft sensor modeling method with dynamic time-delay estimation and its application in wastewater treatment plant

Time-delay estimation is an important step for soft sensor modeling. In practical industrial process, the transportation time of materials and the transmission time of signals are fluctuating, and the traditional static time-delay estimation (STDE) methods can no longer accurately extract the dynamic time-delay characteristics. In this paper, a weighted relevance vector machine model based on dynamic time-delay estimation (DTDE-WRVM) is proposed to realize the dynamic characteristic extraction and quality variable prediction. Firstly, a static delimitation-dynamic update (SD-DU) strategy is adopted for dynamic time-delay estimation, where the dynamic time-delay variation range can be determined based on the fuzzy curve analysis (FCA) method, and the dynamic time-delay characteristics between process variables can be captured based on improved sliding time window. Then, a weighted soft sensor model is established by setting the weights for the variables with different time-delays according to their importance to the output variable. Finally, a numerical example and a wastewater treatment plant (WWTP) case are studied to demonstrate the effectiveness of the DTDE-WRVM model. With the proposed method, the prediction accuracy of WWTP is significantly improved, e.g., the value of the root-mean-square error decreases by 55.7% from 0.0972 to 0.0431.

Session-based news recommendations using SimRank on multi-modal graphs

Recommender systems are among the most widespread applications of artificial intelligence techniques. For instance, news recommender systems serve users in managing the overload of information they come across when accessing news portals. Obviously, in the news domain time-awareness of recommendation approaches are crucial. However, most of these approaches missed to consider user sessions, which group the items that a user interacted with. In this paper, we study the problem of session-based recommendations by running SimRank on time-evolving heterogeneous graphs. In particular, we construct a dynamic heterogeneous multi-partite graph and adjust SimRank to run on it by using different (i) sliding time window sizes, (ii) sub-graphs used for model learning and (iii) sequential article weighting strategies. We evaluate our algorithms on two real-life datasets, and we show that our method outperforms other state-of-the-art methods in terms of accuracy and diversity. The significance and impact of this work is important because it introduces to the research community of expert and intelligent systems, for the first time, a stream-based version of SimRank algorithm, which is able to run over time-evolving graphs.