Annalisa Appice Knihy

Sensor networks comprise of a number of sensors installed across a spatially distributed network, which gather information and periodically feed a central server with the measured data. The server monitors the data, issues possible alarms and computes fast aggregates. As data analysis requests may concern both present and past data, the server is forced to store the entire stream. But the limited storage capacity of a server may reduce the amount of data stored on the disk. One solution is to compute summaries of the data as it arrives, and to use these summaries to interpolate the real data. This work introduces a recently defined spatio-temporal pattern, called trend cluster, to summarize, interpolate and identify anomalies in a sensor network. As an example, the application of trend cluster discovery to monitor the efficiency of photovoltaic power plants is discussed. The work closes with remarks on new possibilities for surveillance enabled by recent developments in sensing technology.

The three volume set LNAI 9284, 9285, and 9286 constitutes the refereed proceedings of the European Conference on Machine Learning and Knowledge Discovery in Databases, ECML PKDD 2015, held in Porto, Portugal, in September 2015. The 131 papers presented in these proceedings were carefully reviewed and selected from a total of 483 submissions. These include 89 research papers, 11 industrial papers, 14 nectar papers, 17 demo papers. They were organized in topical sections named: classification, regression and supervised learning; clustering and unsupervised learning; data preprocessing; data streams and online learning; deep learning; distance and metric learning; large scale learning and big data; matrix and tensor analysis; pattern and sequence mining; preference learning and label ranking; probabilistic, statistical, and graphical approaches; rich data; and social and graphs. Part III is structured in industrial track, nectar track, and demo track.

This book constitutes the thoroughly refereed post-conference proceedings of the Third International Workshop on New Frontiers in Mining Complex Patterns, NFMCP 2014, held in conjunction with ECML-PKDD 2014 in Nancy, France, in September 2014. The 13 revised full papers presented were carefully reviewed and selected from numerous submissions. They illustrate advanced data mining techniques which preserve the informative richness of complex data and allow for efficient and effective identification of complex information units present in such data. The papers are organized in the following classification and regression; clustering; data streams and sequences; applications.

This book constitutes the thoroughly refereed conference proceedings of the First International Workshop on New Frontiers in Mining Complex Patterns, NFMCP 2012, held in conjunction with ECML/PKDD 2012, in Bristol, UK, in September 2012. The 15 revised full papers were carefully reviewed and selected from numerous submissions. The papers are organized in topical sections on mining rich (relational) datasets, mining complex patterns from miscellaneous data, mining complex patterns from trajectory and sequence data, and mining complex patterns from graphs and networks.