News: Ask the Panel

We setup an online tool for asking questions to the panellists. You can add your relevant question or up-vote existing questions that you would like the panellists to answer. The list can be accessed here.

News: Accepted Papers
List of accepted papers is now available here.

Description

Anomaly, outlier and novelty detection methods are crucial tools in any data scientist’s inventory and are critical components of many real-world applications. Abnormal user activities can be used to detect credit card fraud, network intrusions or other security breaches. In computational biology, characterization of systematic anomalies in gene expression can be translated into clinically relevant information. With the rise of Internet-of-Things, the task of monitoring and diagnostics of numerous autonomous systems becomes intractable for a human and needs to be outsources to a machine. Early detection of an upcoming earthquake or tsunami can potentially save human lives. These applications make anomaly detection methods increasingly relevant in the modern world.

However, with the advent of Big Data, new challenges and questions are introduced, which will need to be addressed by the next generation of the anomaly and outlier detection algorithms. The goal of our workshop is to survey the existing techniques and discuss new research directions in this area:

1. How can one detect anomalies in streaming data?
2. How to address non-stationarity in data when the data distribution (and anomalies) is changing with time?
3. How to detect anomalies as early as possible?
4. How to perform “structured anomaly detection”? I.e., how to detect anomalies that are sequences, trees, graphs, or in general, data that violates the IID assumptions?
5. Can weakly labeled or partially labeled data help? If so, then how do we take advantage of this labeled data?
6. How can we deal with huge data sizes?
7. Can we learn meaningful data representations for anomaly detection?
8. How can we accurately evaluate performance in settings with strongly unbalanced datasets or positive (and unlabeled) examples only?
9. Can we perform “significant anomaly detection” (e.g., using p-values)?
10. What is an “anomaly”? How does it compare to outliers, novelties and the like and can they be tackled with the same methodologies?
11. How can we explain decisions of anomaly detectors to guide human experts?


Invited Speakers

Leman Akoglu, Stony Brook University
Thomas Dietterich, Oregon State University
Klaus-Robert Mueller, Berlin Institute of Technology
Clayton Scott, University of Michigan


Schedule

SessionTimeSpeakerTitle





9.00 - 9.15Opening Remarks
19.15 - 10.00Thomas DietterichTBA









10.00 - 10.30Coffee Break & Poster Session





10.30 - 11.15Leman AkogluTackling Anomaly Mining Problems in the Wild
211.15 - 12.00Clayton ScottMixture Proportion Estimation: Estimating the Proportions of Anomalies and Beyond









12.00 - 13.30Lunch





13.30 - 14.15Klaus-Robert MuellerAnomalies: algorithms, applications and perspectives
314.15 - 15.00Panel Discussion









15.00 - 15.30Coffee Break & Poster Session





15.30 - 15.45Contributed Paper: Nicolas GoixHow to Evaluate the Quality of Unsupervised Anomaly Detection Algorithms?
415.45 - 16.00Contributed Paper: Björn BarzMaximally Divergent Intervals for Anomaly Detection

16.00 - 16.15Contributed Paper: Albert ThomasLearning Hyperparameters for Unsupervised Anomaly Detection

16.15 - 16.30Closing and Awards


Call for Papers

We solicit submission of research papers in the area of anomaly, outlier and novelty detection with the focus on the topics outlined above. Relevant papers that have been recently published or presented elsewhere are allowed, provided that previous publications are explicitly acknowledged. Submission must adhere to ICML 2016 style format and max. 4 pages long, including figures (+additional fifth page containing cited references, supplementary material can be provided). As the review process is not blind authors may reveal their identity during submission process. Please submit your manuscripts to icml2016ad@gmail.com.

All accepted papers will have a poster presentation and the best papers will be selected for an oral presentation.



Important dates:

Submission deadline: May 1st
Author notification: May 10th
Camera-ready version due: June 15th


Location and Dates:

This workshop will take place on June 24th after ICML 2016 main conference which is held in New York, USA. The venue is Marriott Marquis hotel located right in the middle of the iconic Times Square in New York City.

Organizers

Nico Goernitz (Berlin Institute of Technology)
Marius Kloft (Humboldt University of Berlin)
Vitaly Kuznetsov (Google Research)



Thank you to our sponsors