CAREER: A General Framework for Methodical and Interpretable Anomaly Mining

PI: Leman Akoglu Phone: 1 (412) 268 3043
H. John Heinz III College 2118C Hamburg Hall
Carnegie Mellon University Email: lakoglu AT
Pittsburgh, PA 15213 Website:

This material is based upon work supported by the National Science Foundation under Grant No. 1452425. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.


1.1. Abstract

Link to NSF abstract

1.2. Keywords

Anomaly Mining, Novel Anomaly Definitions, Descriptive Methods, Anomaly Ensembles, Anomaly Mining Framework, Heterogeneous Data Mining,
Graph Mining, Applications.

1.3. Funding agency


In addition to the PIs, the following graduate students work on the project.


3.1. Project goals

Technical Merits:

This proposal aims to push the boundaries of anomaly mining as a field through a quest for principled foundations and practices. Research will create previously unstudied classes of data representations that unify heterogeneous data sources, and build on them to formulate novel anomaly mining problems. We will invent new, descriptive algorithms for complex anomaly detection and characterization, that will also explore and exploit ensemble and multi-view approaches. The proposed research will give rise to a comprehensive framework for anomaly mining; through a deeper understanding of the space of problems and objectives, new models and algorithms, and systematic techniques to harness them.

Broader Impacts:

Societal: Proposed research will take the essential steps to mature anomaly mining into a valuable contributor to the larger world. It will have direct significance to many concrete problems (e.g., outsider threat, fraud, intrusion) important for the government, industry, and the society. The project will build a web-based platform that hosts a repository of formulations, algorithms, tools, and datasets, for the research community and the public to leverage. We will collaborate with industry and hospital partners to shepherd our innovations into deployed technology, with tangible impact on security and healthcare.
Educational: The PI is committed to developing an education plan that: trains students to think creatively in formulating and solving problems, enhances undergraduate training by involving Honors thesis students in proposed research, promotes campus wide synergism for students across departments, and increases the role of women in Computer Science through mentoring and open house events for women in the community.

The project summary can be found here.

3.2. Publications

3.3. Workshops


The educational contributions of the project include:

Point of Contact: Leman Akoglu, lakoglu AT

Last updated: May 2020, by Leman Akoglu