Outlier mining is a major task in data analysis. Outliers are objects that highly deviate from regular their local neighborhood. Density-based outlier ranking methods score each object based on its degree of deviation. In many applications, these degenerate to random listings due low contrast between outliers and objects. do not show up the scattered full space, they hidden multiple high subspace projections data. Measuring such subspaces for rankings an open research challenge. this work,...

Decentral Smart Grid Control (DSGC) is a new system implementing demand response without significant changes of the infrastructure. It does so by binding electricity price to grid frequency. While models DSGC exist, they rely on various simplifying assumptions. For example, researchers have assumed that behavior all participants in identical. In this paper we study how data-mining techniques can help remove some these simplifications, while keeping representation insights concise. We...

Outlier analysis is an important data mining task that aims to detect unexpected, rare, and suspicious objects. ranking enables enhanced outlier exploration, which assists the user-driven analysis. It overcomes binary detection of outliers vs. regular objects, not adequate for many applications. Traditional techniques focus on either vector or graph structures. However, today's databases store both, multi dimensional numeric information relations between objects in attributed graphs. An open...

Outlier mining is an important task for finding anomalous objects. In practice, however, there not always a clear distinction between outliers and regular objects as have different roles w.r.t. attribute sets. An object may deviate in one subspace, i.e. subset of attributes. And the same might appear perfectly other subspaces. One can think subspaces multiple views on database. Traditional methods consider only view (the full space). Thus, they miss complex that are hidden this work, we...

Current mining algorithms for attributed graphs exploit dependencies between attribute information and edge structure, referred to as homophily. However, techniques fail if this assumption does not hold the full space. In multivariate spaces, some attributes have high dependency with graph structure while others do show any dependency. Hence, it is important select congruent subspaces (i.e., subsets of node attributes) showing structure. work, we propose a method statistical selection such...

Previous chapter Next Full AccessProceedings Proceedings of the 2013 SIAM International Conference on Data Mining (SDM)CMI: An Information-Theoretic Contrast Measure for Enhancing Subspace Cluster and Outlier DetectionHoang Vu Nguyen, Emmanuel Müller, Jilles Vreeken, Fabian Keller, Klemens BöhmHoang Böhmpp.198 - 206Chapter DOI:https://doi.org/10.1137/1.9781611972832.22PDFBibTexSections ToolsAdd to favoritesExport CitationTrack CitationsEmail SectionsAboutAbstract In many real world...

Outlier ranking aims at the distinction between exceptional outliers and regular objects by measuring deviation of individual objects. In graphs with multiple numeric attributes, not all attributes are relevant or show dependencies graph structure. Considering both structure given one cannot measure a clear This is because existence irrelevant clearly hinders detection outliers. Thus, has to select local outlier contexts including only those showing high contrast deviating It an open...

Human Activity Recognition (HAR) from devices like smartphone accelerometers is a fundamental problem in ubiquitous computing. Machine learning based recognition models often perform poorly when applied to new users that were not part of the training data. Previous work has addressed this challenge by personalizing general unique motion pattern user static batch setting. They require target data be available upfront. The more challenging online setting received less attention. No samples are...

Load disaggregation methods infer the energy consumption of individual appliances from their aggregated consumption. This facilitates savings and efficient management. However, most existing work on load has only considered household settings. may be due to companies preferring not share data, rendering such data hardly available.

In this article metadata for mulimedia documents are classified in conformity with their nature, and the different kinds of brought into relation purposes intended. We describe how may be organized accordance ISO standards SGML, which facilitates handling structured documents, DFR, supports storage collections documents. Finally, we outline impact our observations on future developments.

Security mechanisms are essential for business processes. Currently, business-process-management systems (BPMSs) provide relatively little security support, and programmers must be familiar with interfaces of that not part BPMSs. Enforcing constraints a process leads to high implementation maintenance costs. Our approach in turn is wholistic one, providing support from the modelling runtime phase businessprocess lifecycle. As current security-modelling approaches lack features important...

Many renewable sources for electricity generation are distributed and volatile by nature, become inefficient difficult to coordinate with traditional power transmission paths. As a part of the transition from fossil fuel sources, local energy markets allow an efficient allocation distribution nearby households. When using discrete time double auction model, bids in such reflect supply demand energy. However, since household contains personal information, not line privacy legislation. In this...

Benchmarking unsupervised outlier detection is difficult. Outliers are rare, and existing benchmark data contains outliers with various unknown characteristics. Fully synthetic usually consists of regular instance clear characteristics thus allows for a more meaningful evaluation methods in principle. Nonetheless, there have only been few attempts to include benchmarks detection. This might be due the imprecise notion or difficulty arrive at good coverage different domains data. In this work...

Valuation-aware traffic-control mechanisms for road intersections take the valuations of reduced waiting time individual drivers into account. They use agents to avoid any disturbance driver, and they feature specifically designed negotiating valuations. While such do indeed increase overall driver satisfaction, only allow one vehicle at a intersection so far. But concurrent utilization an is important needed in reality. This paper proposes new auction-based intersectioncontrol allowing...

Data-flow errors in BPMN 2.0 process models, such as missing or unused data, lead to undesired executions. In particular, since with a standardized execution semantics allows specifying alternatives for data well optional identifying systematically is difficult. this paper, we propose an approach detecting data-flow models. We formalize models by mapping them Petri Nets and unfolding the regarding data. define set of anti-patterns representing By employing anti-patterns, our tool performs...