spatial-temporal anomaly detection methods are mostly used for single object, but rarely for multiple objects with changing positions. This problem is often encountered in multi-player online battle arena (MOBA) games...
详细信息
spatial-temporal anomaly detection methods are mostly used for single object, but rarely for multiple objects with changing positions. This problem is often encountered in multi-player online battle arena (MOBA) games, train control systems and modern battlefield command systems, and so on. However, due to the time dependence, object correlation and Display Constraint, there are few methods for solving such problem properly. In this paper, we defined the problem of multi-object spatial-temporal anomaly detection with Display Constraint in detail. To address this problem, we proposed a long short-term memory (LSTM)-based framework. First, we proposed a Display Constraint Graph to represent location relationship and designed an LSTM framework to calculate the reconstruction error. Then we used the DCG based anomaly score to discriminate abnormal subsequences and objects. We applied this method to 18 MOBA game data streams, and achieved better results than traditional methods.
Much of the big data which is produced is due to IoT devices and various sensor networks. This data often comes with spatial as well as temporal properties that can tell investigators many things about the environment...
详细信息
ISBN:
(纸本)9781538619568
Much of the big data which is produced is due to IoT devices and various sensor networks. This data often comes with spatial as well as temporal properties that can tell investigators many things about the environment in which they are located. For security practitioners, how to find abnormal activities or anomalies in the vast amount of spatial-temporal dynamic data is a daunting task. We present a system, STAnD, to assist investigators in determining patterns within these spatial-temporal data sets. The analysis conducted by using this program can support correlating events in both the spatial and temporal domains which will lead the investigators to determine probable causes for potential malicious events.
The development of modern manufacturing has raised greater demands on the accuracy, response speed, and operating cost of industrial accident warnings. Compared to conventional contact sensors, surveillance cameras ca...
详细信息
The development of modern manufacturing has raised greater demands on the accuracy, response speed, and operating cost of industrial accident warnings. Compared to conventional contact sensors, surveillance cameras can contactlessly capture spatial-temporal information of the open workspace with stable data quality, widely used in industrial process monitoring. However, due to the scarcity of industrial video datasets and the rarity and diversity of abnormal events, existing video -based anomalydetection models perform poorly in manufacturing scenarios. In this regard, we collect two datasets from typical industrial sites and propose a memory -enhanced spatial-temporal encoding (MSTE) framework for automatic industrial anomalydetection. The proposed MSTE framework learns spatial and temporal normality as well as spatial-temporal correlations with parallel structures and simultaneously measures deviations in appearance, motion, and consistency to respond to complex industrial anomalies accurately. Experimental results on public benchmarks and realworld industrial videos show that our method outperforms existing methods and achieves accurate temporal localization of various spatial-temporal anomalies, which helps to improve the safety and reliability of intelligent manufacturing.
This article deals with the spatial-temporal investigation of the possible relationship of the carbon monoxide (CO) anomalies with the 2013 Mw =6.7 Lushan earthquake. The complexity of the seismogenic environment cont...
详细信息
This article deals with the spatial-temporal investigation of the possible relationship of the carbon monoxide (CO) anomalies with the 2013 Mw =6.7 Lushan earthquake. The complexity of the seismogenic environment contributes to the irregular spatial distribution of CO background. Accurate estimation of the spatial reference background is crucial for precisely detecting seismic-related anomalies. In this study, we proposed an improved Robust Satellite Techniques (RST) approach, in which long-term spatial common variation was incorporated into reference background using multichannel singular spectrum analysis (MSSA) together with spatial clustering. Using the MSSA-RST approach, we detected CO anomalies associated with the Lushan earthquake, utilizing MERRA-2 data spanning from 180 days before to 90 days after the mainshock, and compared the results with those obtained using typical RST. The cumulative CO anomalies exhibited two accelerated growths in the form of a sigmoidal trend: from -51 to -40 days, and from -26 days to 15 days after the mainshock. spatially, CO anomalies were distributed around the epicenter and along fault zones. Subsequently, we analyzed seismic events and deformation to explore the association between CO anomalies and the Lushan earthquake. The lithospheric Benioff strain also displayed two sigmoid accelerations preceding the mainshock, consistent with cumulative CO anomalies. Some high-value regions in the GPS velocity field were aligned with CO anomaly clusters surrounding Longmenshan, Xianshuihe, Zemuhe, and Anninghe faults. Furthermore, the spatial correspondence and temporal sequence of multiple parameters provide additional support for the potential seismic origin of most atmospheric CO anomalies detected by the new approach, which can be explained by lithosphere-atmosphere-ionosphere coupling (LAIC) models.
暂无评论