Brian's Research Page
- Information diffusion and influence in information networks
- developing new streaming models for information networks
- studying patterns of information diffusion
- identifying dependencies between temporal processes
- Anomaly detection in communication networks
- discovering communication patterns in the data
- identifying anomalies based on both temporal information and graph structure
- want efficient solutions that are scalable to large networks
Past Projects:
- Anonymization of recommendation databases, e.g. the Netflix Challenge dataset (published in ASIACCS '10)
- preserving privacy and utility when publishing sparse recommendation datasets
- want to minimize the loss in prediction accuracy
- Privacy-preserving database-as-a-service (published in PETS '09)
- allow service providers hosting an outsourced DB to answer aggregate queries
- individual entries are kept private from both users and the service providers
- users can verify the accuracy of query responses, relying only on trust of data owner
- Anonymization of social network graphs (published in ASIACCS '09)
- want to publish structure of a social network graph (e.g. for informational, corporate, or research use)
- need to protect constituents from re-identification attacks based on knowledge of local graph structure
- how to achieve desired trade-off between privacy goals and utility of published graphs
- "Streaming Models and Algorithms for Communication and Information Networks". Brian Thompson and James Abello. [In preparation] [presentation]
- "Towards Publishing Recommendation Data With Predictive Anonymization". Chih-Cheng Chang, Brian Thompson, Hui (Wendy) Wang, and Danfeng Yao. ACM Symposium on Information, Computer and Communications Security (ASIACCS), April 2010. [paper] [presentation]
- "Privacy-Preserving Computation and Verification of Aggregate Queries on Outsourced Databases". Brian Thompson, Stuart Haber, William G. Horne, Tomas Sander, and Danfeng Yao. Privacy-Enhancing Technologies Symposium (PETS), August 2009. [paper] [presentation]
- "The Union-Split Algorithm and Cluster-Based Anonymization of Social Networks". Brian Thompson and Danfeng Yao. ACM Symposium on Information, Computer and Communications Security (ASIACCS), March 2009. [paper] [presentation]
- "The Early Bird Gets the Buzz: Detecting Anomalies and Emerging Trends in Information Networks". Brian Thompson. Doctoral Consortium at International Conference on Web Search and Data Mining (WSDM), February 2011. [2-page extended abstract] [5-page extended abstract]
- "A Renewal Theory Approach to Anomaly Detection in Communication Networks". Brian Thompson and Tina Eliassi-Rad. Workshop on Information in Networks (WIN), September 2010. [poster] [extended abstract]
- "DAPA-V10: Discovery and Analysis of Patterns and Anomalies in Volatile Time-Evolving Networks". Brian Thompson and Tina Eliassi-Rad. Workshop on Information in Networks (WIN), September 2009. [poster] [extended abstract]
- "Privacy-Aware Publishing of Netflix Data". Brian Thompson, Chih-Cheng Chang, and Danfeng Yao. IEEE Symposium on Security and Privacy, May 2009. [poster]
- "Social Role-Preserving Graph Anonymization Using Clustering". Brian Thompson and Danfeng Yao. IBM Security and Privacy Day, December 2008. [poster]