Detecting outliers in a large set of data objects is major mining task aiming at finding different mechanisms responsible for groups set. All existing approaches, however, are based on an assessment distances (sometimes indirectly by assuming certain distributions) the full-dimensional Euclidean space. In high-dimensional data, these approaches bound to deteriorate due notorious "curse dimensionality". this paper, we propose novel approach named ABOD (Angle-Based Outlier Detection) and some...

The detection performance of small objects in remote sensing images has not been satisfactory compared to large objects, especially low-resolution and noisy images. A generative adversarial network (GAN)-based model called enhanced super-resolution GAN (ESRGAN) showed remarkable image enhancement performance, but reconstructed usually miss high-frequency edge information. Therefore, object degradation for on recovered Inspired by the success (EEGAN) ESRGAN, we applied a new edge-enhanced...

In recent years, the research community introduced various methods for processing skyline queries in multidimensional databases. The operator retrieves all objects being optimal w.r.t. an arbitrary linear weighting of underlying criteria. most prominent example query is to find a reasonable set hotels which are cheap but close beach. this paper, we propose new approach computing skylines on routes (paths) road network considering multiple preferences like distance, driving time, number...

Multiplayer Online Battle Arena (MOBA) games are among the most played digital in world. In these games, teams of players fight against each other arena environments, and gameplay is focussed on tactical combat. this paper, we present three data-driven measures spatio-temporal behaviour Defence Ancients 2 (DotA 2): 1) Zone changes; 2) Distribution team members and: 3) Time series clustering via a fuzzy approach. We method for obtaining accurate positional data from DotA 2. investigate how...

Monitoring tree regeneration in forest areas disturbed by resource extraction is a requirement for sustainably managing the boreal of Alberta, Canada. Small remotely piloted aircraft systems (sRPAS, a.k.a. drones) have potential to decrease cost field surveys drastically, but produce large quantities data that will require specialized processing techniques. In this study, we explored possibility using convolutional neural networks (CNNs) on automatically detecting conifer seedlings along...

In applications of biometric databases the typical task is to identify individuals according features which are not exactly known. Reasons for this inexactness varying measuring techniques or environmental circumstances. Since these circumstances necessarily same when determining different individuals, exactness might strongly vary between as well features. To similarity search on feature vectors applicable, but even use adaptable distance measures capable handle objects having an individual...

The detection performance of small objects in remote sensing images is not satisfactory compared to large objects, especially low-resolution and noisy images. A generative adversarial network (GAN)-based model called enhanced super-resolution GAN (ESRGAN) shows remarkable image enhancement performance, but reconstructed miss high-frequency edge information. Therefore, object degrades for on recovered Inspired by the success (EEGAN) ESRGAN, we apply a new edge-enhanced (EESRGAN) improve...

Integrity management of wind turbine structures depends on effective condition control through inspections and maintenance throughout their service lives. In practice, the integrity strategies often overlook potential information that can be gathered during operation structures, using structural health monitoring (SHM) systems. Utilization SHM as a means to inform optimal risk-based inspection planning is crucial ensure coherent with evidence may established at relatively low costs....

When automatically extracting information from the world wide web, most established methods focus on spotting single HTML-documents. However, problem of complete web sites is not handled adequately yet, in spite its importance for various applications. Therefore, this paper discusses classification sites. First, we point out main differences to page by discussing a very intuitive approach and weaknesses. This treats site as one large HTML-document applies well-known classification. Next,...

In modern application domains such as multimedia, molecular biology and medical imaging, similarity search in database systems is becoming an increasingly important task. Especially for CAD applications, suitable models can help to reduce the cost of developing producing new parts by maximizing reuse existing parts. Most are based on feature vectors. this paper, we shortly review three which pursue paradigm. Based most promising these models, explain how sets vectors be used more effective...

Although in 2014, Switzerland had an average of less than two fatalities per billion vehicle-kilometres, making its roads among the safest Europe, still more 17 000 traffic accidents occurred on Swiss communal roads, cantonal and national highways. On highway network approximately 1800 km alone, there were almost 1700 involving personal injuries. In order to further reduce this number accidents, it is important that accident risks are assessed as accurately possible. A state-of-the-art...

Abstract Colluvial sediments originating from soil erosion on slopes have proven to constitute significant evidence for tracing past human impact mountain landscapes. In the Central European Erzgebirge (Ore) Mountains, colluvial are associated with specific landforms (footslopes, slope flattenings, dells) and cover a share of 11% (11,905 ha) regional landscape. Thirteen pedosedimentary sections layers were investigated at five forested sites (520–730 m a.s.l.) within context mining...

Recent transformer-based offline video instance segmentation (VIS) approaches achieve encouraging results and significantly outperform online approaches. However, their reliance on the whole immense computational complexity caused by full Spatio-temporal attention limit them in real-life applications such as processing lengthy videos. In this paper, we propose a single-stage efficient VIS framework named InstanceFormer, which is especially suitable for long challenging We three novel...

Semantic segmentation represents a fundamental task in computer vision with various application areas such as autonomous driving, medical imaging, or remote sensing. For evaluating and comparing semantic models, the mean intersection over union (mIoU) is currently gold standard. However, while mIoU serves valuable benchmark, it does not offer insights into types of errors incurred by model. Moreover, different may have impacts on downstream applications. To address this issue, we propose an...

The assessment of rock-fall hazards is subject to significant uncertainty, which not fully considered in general practice and research. This paper reviews classifies the various sources uncertainty. Taking a generic framework for risk as source, probabilistic model presented that consistently combines different types uncertainties, order obtain unified estimate risk. An important aspect it allows incorporating all available information, including physical empirical models, observations...

Previous chapter Next Full AccessProceedings Proceedings of the 2008 SIAM International Conference on Data Mining (SDM)Statistical Density Prediction in Traffic NetworksHans-Peter Kriegel, Matthias Renz, Schubert, and Andreas ZuefleHans-Peter Zueflepp.692 - 703Chapter DOI:https://doi.org/10.1137/1.9781611972788.63PDFBibTexSections ToolsAdd to favoritesExport CitationTrack CitationsEmail SectionsAboutAbstract Recently, modern tracking methods started allow capturing position massive numbers...