NASA Logo, National Aeronautics and Space Administration
Intelligent Systems Division Banner

Dr. Kamalika Das Gives Two Invited Talks on Anomaly Detection

Data Sciences Group research scientist Dr. Kamalika Das gave two invited presentations on anomaly detection. The first was an invited lecture in Carnegie Mellon University (CMU) Silicon Valley's graduate Machine Learning course taught by Ole Mengshoel. The second presentation was an invited talk at the 2018 HSBC Collaborate to Innovate (C2I) Analytics Summit. The CMU lecture mainly focused on graph-based anomalies and described work that Kamalika did as part of her 2012 Defense Advanced Research Projects Agency (DARPA) Anomaly Detection at Multiple Scales (ADAMS) grant in collaboration with PARC, and later extended as part of her 2014 NASA Computational Modeling Algorithms and Cyberinfrastructure (CMAC) grant. For the HSBC invited talk, Kamalika covered a broader range of topics within anomaly detection, including the Multiple Kernel Anomaly Detection (MKAD) algorithm that was developed by the Data Sciences group as part of the Aeronautics Research Mission Directorate (ARMD) Integrated Vehicle Health Management IVHM project in 2010.

BACKGROUND: An anomaly is a pattern within data that does not conform to expected behavior. Anomaly detection has its own unique challenges within machine learning that stem from the problem of looking for unknown patterns. At the same time these challenges transcend boundaries of application and are often ubiquitous across domains. That is why algorithms that we have developed for finding anomalies in operational flight data and in social networks can be applied to financial fraud detection and climate networks, respectively. The anomaly detection algorithms described in these talks deal with finding anomalies in heterogeneous time series data sets consisting of real, categorical, and text sources, as well relational anomalies in time-evolving networks.

The Data Sciences (DS) group is a collaboration of scientists researching core data mining and machine-learning algorithms and their applications in a variety of fields, including Integrated System Health Management (ISHM), aeronautics, space and exploration, and Earth sciences to help answer questions that are of relevance to one or more NASA missions.

NASA PROGRAM FUNDING: DARPA Anomaly Detection at Multiple Scales (ADAMS) and NASA Computational Modeling Algorithms and Cyberinfrastructure (CMAC) grants, NASA Aviation Operations and Safety Program (AOSP)

COLLABORATORS: Aniruddha Basak (CMU), Sricharan Kumar (PARC), and Bryan Matthews (DS group, Code TI)

POINT OF CONTACT: Kamalika Das, kamalika.das@nasa.gov

First Gov logo
NASA Logo - nasa.gov