Differences

This shows you the differences between two versions of the page.

--- papers:bakir-bottou-weston-2005 [2006/04/12 19:14]
127.0.0.1 (old revision restored)
+++ papers:bakir-bottou-weston-2005 [2007/07/24 19:11]
leonb
@@ Line 1: / Line 1: @@
 ===== Breaking SVM Complexity with Cross-Training =====
+//Abstract//:
+We propose to selectively remove examples from the training set using
+probabilistic estimates related to editing algorithms
+(Devijver and Kittler, 1982).  This heuristic procedure aims at creating a
+separable distribution of training examples with minimal impact on the
+position of the decision boundary.  It breaks the linear dependency between
+the number of SVs and the number of training examples, and sharply reduces the
+complexity of SVMs during both the training and prediction stages.
 <box 99% orange>
-Gökhan Bakir, Léon Bottou and Jason Weston: Breaking SVM Complexity with Cross-Training,  //Advances in Neural Information Processing Systems//, 17:81-88, Edited by Lawrence Saul, Yair Weiss and Léon Bottou, MIT Press, 2005.
+Gökhan Bakir, Léon Bottou and Jason Weston: **Breaking SVM Complexity with Cross-Training**,  //Advances in Neural Information Processing Systems//, 17:81-88, Edited by Lawrence Saul, Yair Weiss and Léon Bottou, MIT Press, 2005.
 [[http://leon.bottou.org/publications/djvu/nips-2004e.djvu|nips-2004e.djvu]]
@@ Line 11: / Line 18: @@
 </box>
-  @inproceedings{bakir-bottou-weston-2005,
+  @incollection{bakir-bottou-weston-2005,
     author = {Bak{\i}r, G\"{o}khan and Bottou, L\'{e}on and Weston, Jason},
     title = {Breaking SVM Complexity with Cross-Training},
@@ Line 23: / Line 30: @@
   }
+==== Notes ====
+**Much better** solutions for this problem are discussed in
+[[bordes-ertekin-weston-bottou-2005|(Bordes et al., 2006)]]
+and [[collobert-weston-bottou-2006|(Collobert et al., 2006)]].

User Tools

Site Tools

Differences

Page Tools