Adaptive Sliding Window Research Articles

Transformer and Adaptive Threshold Sliding Window for Improving Violence Detection in Videos.

This paper presents a comprehensive approach to detect violent events in videos by combining CrimeNet, a Vision Transformer (ViT) model with structured neural learning and adversarial regularization, with an adaptive threshold sliding window model based on the Transformer architecture. CrimeNet demonstrates exceptional performance on all datasets (XD-Violence, UCF-Crime, NTU-CCTV Fights, UBI-Fights, Real Life Violence Situations, MediEval, RWF-2000, Hockey Fights, Violent Flows, Surveillance Camera Fights, and Movies Fight), achieving high AUC ROC and AUC PR values (up to 99% and 100%, respectively). However, the generalization of CrimeNet to cross-dataset experiments posed some problems, resulting in a 20-30% decrease in performance, for instance, training in UCF-Crime and testing in XD-Violence resulted in 70.20% in AUC ROC. The sliding window model with adaptive thresholding effectively solves these problems by automatically adjusting the violence detection threshold, resulting in a substantial improvement in detection accuracy. By applying the sliding window model as post-processing to CrimeNet results, we were able to improve detection accuracy by 10% to 15% in cross-dataset experiments. Future lines of research include improving generalization, addressing data imbalance, exploring multimodal representations, testing in real-world applications, and extending the approach to complex human interactions.

Adaptive Sliding Window–Dynamic Time Warping-Based Fluctuation Series Prediction for the Capacity of Lithium-Ion Batteries

Accurately predicting the capacity of lithium-ion batteries is crucial for improving battery reliability and preventing potential incidents. Current prediction models for predicting lithium-ion battery capacity fluctuations encounter challenges like inadequate fitting and suboptimal computational efficiency. This study presents a new approach for fluctuation prediction termed ASW-DTW, which integrates Adaptive Sliding Window (ASW) and Dynamic Time Warping (DTW). Initially, this approach leverages Empirical Mode Decomposition (EMD) to preprocess the raw battery capacity data and extract local fluctuation components. Subsequent to this, DTW is employed to forecast the fluctuation sequence through pattern-matching methods. Additionally, to boost model precision and versatility, a feature recognition-based ASW technique is used to determine the optimal window size for the current segment and assist in DTW-based predictions. The study concludes with capacity fluctuation prediction experiments carried out across various lithium-ion battery models. The results demonstrate the efficacy and extensive applicability of the proposed method.

User Behavior Threat Detection Based on Adaptive Sliding Window GAN

User Behavior Threat Detection Based on Adaptive Sliding Window GAN

Vehicle Heading Enhancement Based on Adaptive Sliding Window Factor Graph Optimization for Gyroscope/Magnetometer

Vehicle Heading Enhancement Based on Adaptive Sliding Window Factor Graph Optimization for Gyroscope/Magnetometer

Abnormal Data Detection Based on Adaptive Sliding Window and Weighted Multiscale Local Outlier Factor for Machinery Health Monitoring

Identifying abnormal data to improve data quality is of great importance for machinery health monitoring (MHM). Existing abnormal data detection methods generally depend on appropriate parameter settings and prior knowledge of data distribution, which result in relatively low adaptability to MHM data. To obtain more reliable MHM results, this paper proposed a novel method to detect abnormal data. First, the concept of adaptive sliding window (ASW) is defined and the advantages of ASW in avoiding data leakage and data redundancy are derived. The ASW can optimally divide the overall data into several segments. Next, statistical factors in time- and frequency-domain are extracted from these segments to generate corresponding research objects. Then, an improved weighted multiscale local outlier factor (WMLOF) algorithm is proposed herein to evaluate the data anomaly degree of each object. The WMLOF has the ability to assess and fuse the local outlier factor characteristics at multiple scales, and eventually yields a more comprehensive WMLOF value to evaluate the anomaly degree of the MHM data. Finally, the effectiveness and superiority of the ASW and WMLOF are validated by a synthetic simulation of a faulty rolling bearing and two engineering projects pertaining to a railway vehicle gearbox, and bench test data. Comparative analysis shows that the proposed abnormal data detection method based on the ASW and WMLOF strategies outperforms five reported algorithms in outlier detection.

FAST: A Forecasting Model With Adaptive Sliding Window and Time Locality Integration for Dynamic Cloud Workloads

The workload predictor has attracted attention as a key component of the proactive service operation management framework. However, the request and resource workloads of cloud applications are highly dynamic. Existing approaches decompose the original workload into trend, seasonal, and random components, establish models accordingly, and then combine all outputs to generate results. Indeed, the random component usually has significant heteroscedasticity and noise, having little or even a negative effect on model accuracy improvement. In our model, trend and seasonal components are seen as macro workload changes, and the micro workload changes are obtained by an adaptive sliding window algorithm. Therefore, we propose an ensembling model named FAST for Forecasting workloads with Adaptive Sliding window and Time locality integration. Notably, we propose an adaptive sliding window algorithm that considers trend correlation, time correlation, and random fluctuations of workload for online regression to achieve higher accuracy with lower overhead; and for the error-based integration strategy, we propose a time locality concept for local-predictor behavior and develop a multi-class regression algorithm for model integration. Finally, we conduct experiments on Google cluster trace datasets which show FAST has better accuracy than all other state-of-the-art models for dynamic workloads.

Learning-Based Adaptive Sliding-Window RLNC for High Bandwidth-Delay Product Networks

Learning-Based Adaptive Sliding-Window RLNC for High Bandwidth-Delay Product Networks

Deformation Feature Extraction for GNSS Landslide Monitoring Series Based on Robust Adaptive Sliding-Window Algorithm

Global navigation satellite system technology has been widely used for high-precision, real-time monitoring of landslides. To improve forecasts and early warnings, the true deformation features must be extracted from the global navigation satellite system monitoring series. However, as the deformation rate changes at different creep stages, the relationship between noise and true deformation may also change, making it difficult to accurately describe the deformation. In this study, an adaptive sliding window algorithm is proposed to account for this relationship change. First, the window was defined with an equal window width and step length, which improved the efficiency of feature extraction. Second, the median and normalized interquartile ranges were used to estimate the window samples and obtain a continuous and reliable series. Finally, the window sample breakdown point was defined to adjust the window parameter. These steps were repeated for the adjusted window to achieve adaptive processing of the monitoring series. The results based on both simulated and real landslide monitoring series demonstrated that the proposed method can provide adaptive, robust, and reliable deformation information for landslide warnings. The adaptive sliding window method also successfully assisted in the early warning of a loess landslide in Heifangtai, Gansu province, northwest of the Chinese Loess Plateau, indicating its practical application potential.

Characteristic wavelengths selection of rice spectrum based on adaptive sliding window permutation entropy

Due to the redundancy of rice spectral wavelengths and the strong correlation between adjacent wavelengths, the modeling classification accuracy based on traditional characteristic wavelengths selection methods is insufficient. Thus, a rice spectral characteristic wavelengths selection method based on adaptive sliding window permutation entropy (ASW-PE) was proposed in this paper. Firstly, the ASW-PE algorithm is constructed by combining the adaptive sliding window (ASW) method and permutation entropy (PE) method. Then, for the spectral data of rice varieties WC, XS, YS and YG, based on ASW-PE, sliding window permutation entropy (SW-PE), analysis of variance (ANOVA), competitive adaptive reweighted sampling (CARS) and successive projections algorithm (SPA) to carry out the characteristic wavelengths selection experiment, and evaluated the computational efficiency of the five algorithms from the perspective of time complexity. Finally, a partial least squares (PLS) rice varieties classification model was established based on the spectral characteristic wavelengths selected by the above algorithms, and the characteristic selection performance of the five algorithms was evaluated with the classification accuracy. Experimental results show that the ASW-PE algorithm has a speed advantage in selecting characteristic wavelengths for large sample spectral data. Compared with SW-PE, ANOVA, CARS and SPA algorithms, the accuracy of modeling classification based on ASW-PE method is improved by 5.6%, 22.6%, 8.6% and 15.2%, respectively.

An adaptive IoT architecture using combination of concept-drift and dynamic software product line engineering

Internet of things (IoT) architecture needs to adapt autonomously to the environment and operational to maintain their supreme services. One common problem in the IoT architecture is to manage the reliability of data services, such as sensors’ data, that only sending data to the collector via gateway. If there is a disruption of services, then it is not easy to manage the system reliability. To this, an adaptive environment which is based on software reconfiguration creates a great challenge to provide better services. In this work, the software product line engineering (SPLE) reconfigures the edge devices via rules and software architecture. To identify disruption of data services which can be detected based on anomaly and truncated data. Our work makes use of concept drift to provide a recommendation to the system manager. This is important to avoid misconfiguration in the system We demonstrate our method using an open-source internet of things portal system that integrated to a cluster of sensors which is attached to specific gateway before the data are collected into a cloud storage for further processes. In identifying drifting data, the adaptive sliding window (ADWIN) method outperforms the Page-Hinkley (PH) with more selective identification and sensitive reading.

Deep learning framework for handling concept drift and class imbalanced complex decision-making on streaming data

AbstractIn present times, data science become popular to support and improve decision-making process. Due to the accessibility of a wide application perspective of data streaming, class imbalance and concept drifting become crucial learning problems. The advent of deep learning (DL) models finds useful for the classification of concept drift in data streaming applications. This paper presents an effective class imbalance with concept drift detection (CIDD) using Adadelta optimizer-based deep neural networks (ADODNN), named CIDD-ADODNN model for the classification of highly imbalanced streaming data. The presented model involves four processes namely preprocessing, class imbalance handling, concept drift detection, and classification. The proposed model uses adaptive synthetic (ADASYN) technique for handling class imbalance data, which utilizes a weighted distribution for diverse minority class examples based on the level of difficulty in learning. Next, a drift detection technique called adaptive sliding window (ADWIN) is employed to detect the existence of the concept drift. Besides, ADODNN model is utilized for the classification processes. For increasing the classifier performance of the DNN model, ADO-based hyperparameter tuning process takes place to determine the optimal parameters of the DNN model. The performance of the presented model is evaluated using three streaming datasets namely intrusion detection (NSL KDDCup) dataset, Spam dataset, and Chess dataset. A detailed comparative results analysis takes place and the simulation results verified the superior performance of the presented model by obtaining a maximum accuracy of 0.9592, 0.9320, and 0.7646 on the applied KDDCup, Spam, and Chess dataset, respectively.

Shot Interference Detection and Mitigation for Underwater Acoustic Communication Systems

Underwater acoustic communications (UACs) are demanded in a wide range of marine applications. However, one of the main challenges for UACs is unexpected interferences from nearby acoustic activities, especially dynamic and short-period interferences (called shot interference), which are hard to detect, estimate, and mitigate over time- and frequency-selective acoustic channels. In this article, we propose a novel filter that mitigates the interference effects for both single-carrier and multi-carrier systems without any prior knowledge of the interference. An Adaptive Sliding Window Interference Detection (ASWID) algorithm is developed to detect the location and the number of interfered symbols. Meanwhile, a least trimmed squares (LTS) equalizer is presented for the shot interference estimation and mitigation based on robust regression. The proposed algorithms are evaluated through numerical simulations and real channel measurements. Our results show that the proposed designs can detect and mitigate shot interferences effectively for single-carrier and multi-carrier UAC systems.

Video Summarization Based on Mutual Information and Entropy Sliding Window Method.

This paper proposes a video summarization algorithm called the Mutual Information and Entropy based adaptive Sliding Window (MIESW) method, which is specifically for the static summary of gesture videos. Considering that gesture videos usually have uncertain transition postures and unclear movement boundaries or inexplicable frames, we propose a three-step method where the first step involves browsing a video, the second step applies the MIESW method to select candidate key frames, and the third step removes most redundant key frames. In detail, the first step is to convert the video into a sequence of frames and adjust the size of the frames. In the second step, a key frame extraction algorithm named MIESW is executed. The inter-frame mutual information value is used as a metric to adaptively adjust the size of the sliding window to group similar content of the video. Then, based on the entropy value of the frame and the average mutual information value of the frame group, the threshold method is applied to optimize the grouping, and the key frames are extracted. In the third step, speeded up robust features (SURF) analysis is performed to eliminate redundant frames in these candidate key frames. The calculation of Precision, Recall, and F are optimized from the perspective of practicality and feasibility. Experiments demonstrate that key frames extracted using our method provide high-quality video summaries and basically cover the main content of the gesture video.

Novel hybrid pair recommendations based on a large-scale comparative study of concept drift detection

During the classification of streaming data, changes in the underlying distribution make formerly learned models insecure and imprecise, which is known as the concept drift phenomenon. Online learning derives information from a vast volume of stream data, which are usually affected by these changes in unforeseen ways and are currently generated primarily by the Internet of Things, social media applications, and the stock market. There is abundant literature focused on addressing concept drift using detectors, which essentially attempt to forecast the position of the change to improve the overall accuracy by altering the base learner. This paper presents novel hybrid pairs (classifier and detector) collected from a large-scale comparison of 15 drift detectors; drift detection method (DDM), early drift detection method (EDDM), EWMA for concept drift detection (ECDD), adaptive sliding window (ADWIN), geometrical moving average (GMA), drift detection methods based on Hoeffding’s bound (HDDMA and HDDMW), Fisher exact test drift detector (FTDD), fast Hoeffding drift detection method (FHDDM), Page–Hinkley test (PH), reactive drift detection method (RDDM), SEED, statistical test of equal proportions (STEPD), SeqDrift2, and Wilcoxon rank-sum test drift detector (WSTD) and six classifiers; Naïve Bayes (NB), Hoeffding tree (HT), Hoeffding option tree (HOT), Perceptron (P), decision stump (DS), and k-nearest neighbour (KNN), to determine and recommend the best pair in accordance with the properties of the dataset. The objective of this study is to assess the contribution of a detector to a classifier and obtain the most efficient matched pairs. Through these pairwise comparison experiments, the accuracy rates and evaluation times of the pairs, as well as their false positives, true negatives, false negatives, true positives, drift detection delay, and the MCC. Additionally, the Nemenyi test is employed to compare the pairs against other methods to identify the method(s) for which there is a statistical difference. The results of the experiments indicate that the most efficient pairs—which differed for each dataset type and size—primarily include the HDDMA, RDDM, WSTD, and FHDDM detectors.

A parameter estimation algorithm for multiple frequency-hopping signals based on compressed sensing

In this paper, a parameter estimation algorithm for wideband multiple FH (multi-FH) signals based on compressed sensing (CS) is proposed. Our proposed algorithm is aiming at the condition of existing synchronous and asynchronous frequency-hopping (FH) signals, and meanwhile considering the frequency switching time. Firstly, by segmenting the received signals into several compressed measurement signals, the sparsity is estimated through the compressed spectrum sensing algorithm without reconstructing signals. Secondly, the improved orthogonal matching pursuit (OMP) algorithm is used to the roughly estimate the frequency of multi-FH signals, and the frequency is accurately estimated by frequency clustering algorithm, then the hopping time and hopping speed are roughly estimated through time–frequency analysis. Finally, inspired by the traditional sliding window algorithm, the adaptive sliding window (ASW) algorithm is proposed, which can adaptively adjust the window length and step size so as to increase the accuracy of the hopping time and hopping speed estimation. Numerical simulations demonstrate that the algorithm proposed in this paper can precisely estimate the parameters of multi-FH signals at low signal-to-noise ratio (SNR) and can significantly reduce the algorithm complexity.

A Novel Method of Segmentation and Classification for Meditation in Health Care Systems.

Meditation improves positivity in behavioral as well as psychological changes, which are brought elucidated by knowing neuro-physiological consequences of meditation. In the field of cognitive science, neuroscience and physiological research, Electroencephalogram (EEG) is extensively utilized. The primary tasks of EEG signal analysis is to identify the noisy signal as well as enormous data that create signal processing and subsequent analysis. Beforehand any analysis of the EEG signal, the obtained raw signal must be preprocessed for eliminating undesirable artifacts as well as horrible noise. With the aim of resolving this issue, in this research, raw signals are preprocessed with the help of Band-Pass Filter (BPF) for noise removal method. Instead, in adaptive Sliding Window with Fuzzy C Means Clustering (SW-FCM) segmentation is presented, which precisely as well as automatically segments the signals. So as to analyze the accuracy, five features such as electroencephalography alpha spectrum, frequency of the main peak, Amplitude of the main peak, Higher Order Crossing (HOC), and wavelet features are used as the evaluating variables. Lastly to assess the meditation experience with Fuzzy Kernel least square Support Vector Machine (FKLSSVM) classifier, the presented method with a cross-sectional analysis is utilized. These two classifiers are utilized for meditation experience classification by utilizing an individual feature vector values from equivalent EEG signals. The dataset samples are gathered from Open source Brain-Computer Interface (Open BCI) platform. Outcomes attained are matched up for diverse techniques for identifying as well as for classifying signal segments features using MATLAB. Presented classifiers of the meditation process validate quick interpretation methods that differentiate meditation experience and valuable performance related to artificial approaches for the big-data analysis.

An Adaptive Sliding-Window Strategy for Outlier Detection in Wireless Sensor Networks for Smart Port Construction

ABSTRACT Yao, H.; Yang, Y.; Fu, X., and Mi, C., 2018. An adaptive sliding-window strategy for outlier detection in wireless sensor networks for smart port construction. In: Ashraf, M.A. and Chowdhu...

Adaptive spectrum scanning techniques for reducing the identification time of the frequency coded chipless RFID system

AbstractThe main objective of this contribution is to introduce novel techniques for reducing the time taken from the reader to identify the frequency coded chipless radio frequency identification tags existed in the reader's interrogation region, system latency. The frequency scanning methodology, number of averaging for clutter removal, and hop duration are the 3 main parameters that significantly affect the overall system latency. Consequently, the adaptive frequency hopping (AFH) and adaptive sliding window (ASW) methodologies are proposed and proofed to be efficient for the chipless radio frequency identification systems from the latency and accuracy perspectives. Likewise, the performance of the designed AFH and ASW techniques are compared with the classical fixed frequency hopping methodology with a fine frequency step to validate the accuracy of the proposed methods. Moreover, 4 different coded frequency coded chipless tags are manufactured and used in the measurements. A real‐world testbed is designed including a software‐defined radio platform by which the proposed adaptive algorithms and traditional fixed frequency hopping methodology are implemented. All the measurements are performed in an indoor realized scenario, including the environmental effects. The experiments show that the proposed AFH combined with ASW algorithms significantly reduce the system latency by 58%.

Adaptive Sliding-Window Strategy for Vehicle Detection in Highway Environments

In highway traffic monitoring systems, vehicle detection is one of the most important tasks. To automatically detect vehicles in general CCTV environments, we should effectively handle the problems caused by the following three issues: severe variation in appearance, ambiguities in location, and ambiguities in size. Over the last decade, these issues have been mainly addressed by employing a sliding-window-based image search. Although it is quite effective in many applications, the manual procedure of initializing necessary size models and high computational costs impede the automation of monitoring systems. To remedy these drawbacks, we propose an adaptive sliding-window strategy, where useful size templates are continuously modeled for a given scene, and the detection is performed by adaptively deforming a sliding window based on the obtained templates. This strategy allows us not only to avoid complex user interferences but also to reduce computational costs for vehicle detection. Experiments on real-world highway environments show that the proposed strategy achieved about 23% improved detection accuracy with 43% reduced processing time over the state-of-the-art sliding-window approach without any user assistance.

A Novel Efficient Adaptive Sliding Window Model for Week-ahead Price Forecasting

In order to improve the accuracy of price forecasting by Web extracting, a novel efficient improved Adaptive Sliding Window (ASW) that the coefficients of the window width can be auto adjusts is proposed in this paper. Agricultural products price based on ASW is utilized to verify validity of adaptive Back Propagation (BP) neural network and adaptive Radial Basis Function (RBF) neural network model respectively. Experiments demonstrated that the Mean Absolute Error (MAE) on ASW model can be getting 99.62 percent accuracy rate. Experiment results proved that the proposed ASW model and adaptive BP neural network model are meaningful and useful to analyze and to research products market, but the proposed ASW model is the best one because of its speed is the fast one which can save time 80 percent than the adaptive BP neural network. DOI : http://dx.doi.org/10.11591/telkomnika.v12i3.4490 Full Text: PDF