Adaptive Techniques for Device Monitoring
We investigate machine learning techniques for detecting any unusual
functioning of a space shuttle component, such as a fuel valve.
Machine learning techniques are especially useful to generate
detection knowledge from historical data. During the monitoring
process, behavior that significantly deviates from the learned model
could indicate potential problems. Our algorithms can significantly
reduce the amount of time and effort to extract, encode, and update
knowledge from experts into monitoring systems.
Data
Commercialization
Publications
- Modeling Multiple Time Series for Anomaly Detection,
P. Chan & M. Mahoney,
Proc. IEEE Intl. Conf. on Data Mining,
p. 90-97, 2005.
- Trajectory Boundary Modeling of Time Series for Anomaly
Detection,
M. Mahoney and P. Chan,
Workshop on Data Mining Methods for Anomaly Detection,
SIGKDD Conf., 2005.
- Learning States and Rules for Detecting Anomalies in Time Series
S. Salvador & P. Chan,
Applied Intelligence, 23(3):241-255, 2005.
- Learning States
for Detecting Anomalies in Time Series
S. Salvador,
MS Thesis, Florida Tech, 2004.
- Determining the Number of Clusters/Segments in Hierarchical Clustering/Segmentation Algorithms,
S. Salvador & P. Chan,
Proc. 16th IEEE Intl. Conf. on Tools with AI, pp. 576-584, 2004.
- FastDTW: Toward Accurate Dynamic Time Warping in Linear Time and Space
S. Salvador & P. Chan,
KDD Workshop on Mining Temporal and Sequential Data, 2004.
- Learning States and Rules for Time Series Anomaly Detection,
S. Salvador, P. Chan & J. Brodie,
Proc. 17th Intl. FLAIRS Conf., pp. 300-305, 2004.
People
- Philip Chan
- Matt Mahoney
- Chirs Tanner
- Stan Salvador (past member)
- John Brodie (past member)
Collaborator
Sponsor
National Aeronautics and Space Administration (NASA)
Related Work
Last modified: Sat Aug 27 18:31:38 EDT 2005