Welcome to the online home of the Event and Pattern Detection Laboratory (EPD Lab) at Carnegie Mellon University’s Heinz College!
Our research focuses on the development of new statistical and computational techniques for scalable and accurate detection of emerging events and other patterns in complex, massive, and high-dimensional datasets. We are particularly interested in applications of this work for the public good, ranging from public health and patient care, to law enforcement and urban analytics, to human rights and conflict.
Our work has advanced the state of the art in large-scale pattern detection in multiple ways, including new statistical methods (which enable more accurate detection of anomalous patterns by integrating multiple data sources, incorporating spatial and temporal information, and using prior knowledge of a domain), new algorithms and data structures (which make previously impossible detection tasks computationally feasible and fast), and new machine learning methods (which enable systems to learn from user feedback, modeling and distinguishing between relevant and irrelevant types of anomaly). We have applied these novel methods to create, develop, and deploy systems that directly benefit the public good, for example, by detecting emerging outbreaks of disease, predicting violent crime hot-spots, and using 311 call data to predict and prevent rodent infestations in Chicago.
More details on our ongoing projects, including both new methods for event and pattern detection and the applications of these methods, are available on our Projects page.