ECCE 2016 - Workshop on Early Classification of Complex Events

Early prediction of temporal events is crucial in many domains including, but not limited to, health informatics, social media, crime analysis, proactive cyber defense, and educational analytics. For example, in health informatics the health of the patient is measured over time and the diagnosis is predicted early, such that intervention is more beneficial; in social media the task is to predict events such as election campaigns as early as possible; in crime analysis the task is to predict crime as early as possible, preferably before it happens; in cyber events where the objective is to detect security threats before it happens; and in educational analytics, the progress of students is used to identify whether the student is going to drop out, such that preventative academic intervention is taking place at the right time.
The emergence of high-throughput technologies and online social network sites challenges researchers to exploit these temporal data in order to identify events of interested as early as possible. Exploiting all the information and structure among actors (e.g. students and social network users) for early prediction of time series is very challenging. How does one provide such prediction in a reasonable time? For some applications, the requirements are even more challenging, such as providing an interpretable model.
Description
In the last few years, the number of publications related to early prediction of temporal events has significantly increased, indicating that there is an urgent need to gather practitioners from different fields to discuss the potentials and challenges related to that topic. Surprisingly, there is no workshop yet designated entirely to discussing this emerging topic. Therefore, the goal of this workshop is to focus on challenges of data mining techniques for early prediction of temporal events. We will bring researchers and experts from both academia and industry to discuss the current state-of-the-art in early prediction of temporal events.
Topics of Interest
Early classification of (multivariate) time series
Early classification of structured data
Early detection of malicious cyber events
Early classification of evolving networks
Cost-sensitive early classification
Domain knowledge constrained early prediction
Interpretable models for early classification
Early classification applications
Workshop Co-Chairs
Mohamed Ghalwash, Temple University
Zoran Obradovic, Temple University
Jian Pei, Simon Fraser University
Program Committee
Hyrum Anderson, Endgame Inc.
Debasish Das, Verizon Wireless
Boris Delibašić, University of Belgrade
Nemanja Djuric, Yahoo Labs
Ahmed Hassan, Ain Shams University
Milos Hauskrecht, University of Pittsburgh
Nathan Parrish, Johns Hopkins University
Vladan Radosavljevic, Yahoo Labs
Gregor Stiglic, University of Maribor
Milan Vukicevic, University of Belgrade
Fei Wang, University of Connecticut
Ping Zhang, IBM T.J. Watson Research Center
Important Dates
Paper Submission: January 20, 2016
Notification of Acceptance: February 5, 2016
Camera Ready Paper Due: February 12, 2016