Learning Networks of People and Places from Location Data
Event details
| Date | 26.06.2009 |
| Hour | 10:15 |
| Speaker | Prof. Tony Jebara, Columbia University, New York |
| Location | |
| Category | Conferences - Seminars |
Networks and graphs have become essential for understanding the online world with applications ranging from the Web to FaceBook. I will discuss building such networks in the offline real world by using location and GPS data. By gathering long-term high frequency location data from millions of mobile devices it becomes possible to track movement trends in real-time in cities, learn networks of real places and learn real social networks of people. We build graphs from this data using generalized matching algorithms and also apply novel visualization, clustering and classification tools to them. For example, we can visualize the network of places in a city showing the similarity between different locations and how active they are right now. Another graph is the network of users showing how similar person X is to person Y by comparing their movement histories and how often they colocated. Embedding and clustering these graphs reveals interesting trends in behavior and organizes people into tribes that are more detailed than traditional demographics. With learning algorithms applied to these human activity graphs, it becomes possible to make predictions for advertising, marketing and collaborative recommendation from real offline behavior.
Prof Jebara's homepage
Practical information
- General public
- Free