Ensemble learning methods: random forests, boosting, PAQ, stacking, and dropout
Despite widespread interest and practical use, the theoretical properties of random forests are still not well understood. One objective of this project is advance the theory of random forests and other ensemble methods, such as PAQ.
In addition to advancing the theory, we are also interested in developing new ensemble methods for streaming data, as well as new scalable esembles with good predictive properties.
Selected Publications
A Machine Learning Perspective on Predictive Coding with PAQ8 Byron Knoll and Nando de Freitas In Data Compression Conference (DCC). Pages 377–386. 2012. |
Consistency of Online Random Forests Misha Denil‚ David Matheson and Nando de Freitas In International Conference on Machine Learning (ICML). Pages 1256–−1264. 2013. JMLR &CPW 28 (3): 1256–1264‚ 2013 |
Info
Principal Investigator |
|
Themes |