Overview of our domain adaptation framework

Project summary

Social media platforms such as Twitter provide valuable information for improving situation awareness and aiding disaster response during emergency events. Machine learning could be used to identify such information. However, supervised learning algorithms rely on labeled data, which is not readily available for a current target disaster. While labeled data might be available for a prior source disaster, supervised classifiers learned from source only may not perform well on the target data, as they don’t make use of target specific features. To address this limitation, our work is focused on domain adaptation approaches, which learn classifiers from unlabeled target data, in addition to source labeled data. Experimental results on the task of identifying tweets relevant to a disaster of interest show that the domain adaptation classifiers can significantly improve the accuracy of the supervised classifiers learned from source data only, in some cases by more than 15%.

Invited Talks

D. Caragea. “Mining Twitter to Aid Disaster Response” - Invited talk at the Kansas City Machine Learning Group monthly meeting (February 2017). slides

D. Caragea. “Learning Domain Adaptation Classifiers from Multiple Distributed Sources” - Invited talk in the session on “Inference and prediction for distributed data” at the 2016 Annual Meeting of the Statistical Society of Canada (June 2016). slides

D. Caragea. “Mining Twitter to Aid Disaster Response” - Invited talk at the Reunion and Seminar “Honoring Beth” (April 2016). slides

Publications

Li, H., Caragea, D. and Caragea, C. (2017). Towards Practical Usage of a Domain Adaptation Algorithm in the First Hours of a Disaster. In: Proceedings of the 14th International Conference on Information Systems for Crisis Response and Management (ISCRAM 2017), France.

Sopova, O., Li, H., and Caragea, D. (2017). Comparison of Two Domain Adaptation Approaches for Classifying Disaster-related Twitter Data In: Proceedings of The 4th International Symposium on Social Networks Analysis, Management and Security (SNAMS 2017), part of the Proceedings of the IEEE International Conference on Future Internet of Things and Cloud (FiCloud-2017), Prague, Czech Republic, 2017.

Li, H., Caragea, D., Caragea, C. and Herndon, N. (2017). Disaster Response Aided by Tweet Classification with a Domain Adaptation Approach. In: Journal of Contingencies and Crisis Management (JCCM), Special Issue on HCI in Critical Systems. In press.

Li, H., Guevara, N., Herndon, N. Caragea, D., Neppalli, K., Caragea, C., Squicciarini, A. and Tapia, A. (2015). Twitter Mining for Disaster Response: A Domain Adaptation Approach. In: 12th International Conference on Information Systems for Crisis Response and Management (ISCRAM 2015), Norway.

Neppalli, K., Caragea, C., Caragea, D., Medeiros, M.C., Tapia, A. and Halse, S. (2017).Predicting Tweet Retweetability during Hurricane Disasters. International Journal of Information Systems for Crisis Response and Management (IJISCRAM). In press.

Neppalli, K., Cerqueira Medeiros, M., Caragea, C., Caragea, D., Tapia, A., Halse, S. (2016). Retweetability Analysis and Prediction during Hurricane Sandy. In: Proceedings of the 13th International Conference on Information Systems for Crisis Response and Management (ISCRAM 2016), Brazil.

disasters.txt · Last modified: 2017/08/25 22:44 by dcaragea
CC Attribution-Noncommercial-Share Alike 4.0 International
www.chimeric.de Valid CSS Driven by DokuWiki do yourself a favour and use a real browser - get firefox!! Recent changes RSS feed Valid XHTML 1.0