Close

Distributed Anomaly Detection for Vertically Partitioned Data Distribution: A Case Study Using Satellite Data from Multiple Modalities

Related Research Areas
Climate Variability & Change, Earth Surface & Interior
Project Description
This project addresses the issue of anomaly detection from distributed databases when different features are observed and stored at geographically different locations. The proposed algorithm uses an one-class SVM based anomaly detection approach for identifying the top $k$ anomalies from the set of observations without centralizing the data to a single location.
Project Administrator(s):
Kamalika Das

Members

foo.jpg
Kamalika Das
Kanishka Bhaduri
foo.jpg
Petr Votava
Matthew Helt
foo.jpg
Armando Rodriguez

Tags

distributed