A common task in many domains involves monitoring routinely collected data for anomalous events. Typically, this inspection of data is performed for surveillance purposes. For instance, a security guard needs to examine video footage for any signs of an intrusion. Scientists also perform a similar task when analyzing experimental data. Detection of anomalous events in experimental data often leads to new scientific discoveries. In order to refer to this monitoring process in as general a term as possible, we will call it event detection. Event detection has the potential to impact a wide range of important real-world applications, ranging from security, finance, public health, medicine, biology, environmental science, manufacturing, astrophysics, business and economics.
In the recent past, human beings have had the laborious job of manually examining collected data for event detection; however, the emergence of computers and massive world interconnectivity have made it easier to collect data and have provided more reasons to do so. Simple forms of data, such as a univariate time series, can be effectively monitored using well-established techniques such as regression, Box-Jenkins models and methods in statistical quality control. Data, however, has become increasingly complex in recent years. Multivariate records, images, video footage, audio recordings, spatial and spatio-temporal data, text documents, and even relational data are now routinely collected. One might expect that existing work in machine learning would be well-suited for this task. However, in practice, the peculiarities of the application often grossly violate the standard assumptions of machine learning. Often, none of the standard paradigms of supervised learning, unsupervised learning or even semi-supervised and active learning fit this situation well. As a result, new algorithms need to be created in order to address these issues and fill an important gap in machine learning research which would impact many of the most pressing real-world applications being studied today.
The presence of event detection problems seems to be widespread throughout areas of machine learning. This workshop will provide a forum for participants from many different communities to share their ideas and experiences. In addition, the workshop will benefit from the diversity of ICML attendees, who span the entire spectrum from applied to theoretical research. The topics of interest include, but are not limited to:
This workshop is intended to be a continuation of the Workshop on Data Mining Methods for Anomaly Detection at KDD 2005 but with a greater emphasis on event detection in richer data representations such as spatio-temporal data, relational data, and unstructured data such as multimedia files.
The workshop will consist of a combination of invited talks, presentations of papers, and a discussion period. Each presentation will be allocated 10-20 minutes for presentation and 10 minutes for discussion. Attendance for the workshop will be limited to 40-50 people.
To participate in the workshop, please send an email message to Weng-Keen Wong (firstname.lastname@example.org) giving your name, affiliation, address, email address, and a brief description of your reasons for wanting to attend. The webpage for the workshop is at