This course is intended for Ph.D. students in Heinz College, the School of Computer Science, and other university departments who wish to engage in cutting-edge research in Event and Pattern Detection. Qualified master's students may also enroll with permission of the instructor; all students are expected to have some prior background in machine learning and/or artificial intelligence (10-601, 10-701, 90-866, or a similar course). We will explore state-of-the-art methods for detection of emerging events and other relevant patterns in massive, high-dimensional datasets, and discuss how such methods can be applied usefully for the public good in medicine, public health, law enforcement, security, and other domains. The course will consist of lectures, discussions on current research articles and future directions, and course projects. Specific topics to be covered may include: anomaly detection, change-point detection, time series monitoring, spatial and space-time scan statistics, pattern detection in graph data, submodularity and LTSS properties for efficient pattern detection, combining multiple data sources, scaling up pattern detection to massive datasets, applications to public health, law enforcement, homeland security, and health care.
Six of the fourteen course meetings will be devoted to discussion of specific topics in Event and Pattern Detection. Each student is expected to give a high quality, twenty-minute PowerPoint presentation, followed by twenty minutes of class discussion, at two of these eight meetings. The primary goals of each topic presentation should be to 1) synthesize and summarize the current state of the art for the given topic, overviewing the major methodological approaches and open problems, 2) to briefly review at least three specific research articles and their relevance to the topic, and 3) to facilitate the remainder of the discussion by posing questions for discussion, preliminary conclusions, and ideas to explore. The chosen research articles should present detection methods and approaches that are new (or not commonly known), or novel applications of event and pattern detection to policy-relevant topics.
To ensure that presentations will be useful and relevant for the class, the presenter(s) should send the instructor a brief text outline of the main topics/points that their presentation will cover, and a proposed set of 1-2 electronically available research articles that the class should read, at least one week prior to the presentation. The assigned reading(s) could be a review paper on the topic, or a landmark work representing a major advance on the topic; it may or may not be one of the three specific articles reviewed in the presentation. The instructor will provide feedback and suggestions, and will post the articles on Blackboard so that the class can read them in advance of the presentation. Some good ways to find relevant papers include looking through recent ML conference proceedings (KDD, ICML, AAAI, NIPS) and journals (MLJ, JMLR, JASA). For many topics, the instructor can suggest a few papers to get you started, and online resources such as Citeseer and Google Scholar will also be helpful.
A list of suggested topics for your presentations, along with some basic information and references about each topic, has been provided in the Course Documents section.
(T 1/12) Course Introduction
Introductions (be prepared to speak for 2-3 minutes each about your background and interests)
Discussion of the course syllabus (course structure, goals, topic presentations, course projects)
Brief lecture/discussion introducing event and pattern detection topics and applications
(H 1/14) Lecture: Individual-Record Anomaly Detection
(T 1/19) Lecture: Purely Temporal Event Detection
(H 1/21) Lecture: Spatial and Space-Time Event Detection
(T 1/26) Lecture: Spatial and Space-Time Event Detection, continued
(H 1/28) Project Proposal
Each student will present a short PowerPoint presentation on their proposed course project, as well as turning in a short (1-2 page) proposal. Please plan to speak for no more than 10 minutes, and leave five minutes for discussion/suggestions from the class.
(T 2/2) Discussion Topics 1-2
Spatial scan methods for detecting irregularly-shaped clusters
Detecting anomalous groups in categorical datasets
2/4) Discussion Topics 3-4
Event detection from text data (news and blogs)
Real-time anomaly detection from streaming data
(H 2/11) Discussion Topics 5-6
Anomaly detection on graphs
Pattern detection on graphs
2/16) Discussion Topics 7-8
(H 2/18) Discussion Topics
Distributed anomaly detection with sensor networks
Event detection using GPS location data
Discussion Topics 11-12
Active learning for rare category detection
Detection using dynamic data
Each student will give a short PowerPoint presentation on their course project. Please plan to speak for no more than 10 minutes, and leave five minutes for questions from the class.
(T 3/2) Final Project Reports due 11:59pm (no class today)