9 search results - page 2 / 2 » Boosting the margin: A new explanation for the effectiveness... |
ICMLA
2004
13 years 6 months ago
2004
AdaBoost rarely suffers from overfitting problems in low noise data cases. However, recent studies with highly noisy patterns clearly showed that overfitting can occur. A natural s...
KDD
2006
ACM
14 years 5 months ago
2006
ACM
To learn concepts over massive data streams, it is essential to design inference and learning methods that operate in real time with limited memory. Online learning methods such a...
CIKM
2005
Springer
13 years 7 months ago
2005
Springer
This paper presents a new pooling method for constructing the assessment sets used in the evaluation of retrieval systems. Our proposal is based on RankBoost, a machine learning v...
TRECVID
2008
13 years 6 months ago
2008
The Informedia team participated in the tasks of Rushes summarization, high-level feature extraction and event detection in surveillance video. For the rushes summarization, our b...