159 search results - page 4 / 32 » A sufficient condition for pooling data |
ICPR
2006
IEEE
14 years 8 months ago
2006
IEEE
It is possible to broadly characterize two approaches to probabilistic modeling in terms of generative and discriminative methods. Provided with sufficient training data the discr...
WWW
2003
ACM
14 years 8 months ago
2003
ACM
In traditional peer-to-peer search networks, operations focus on properly labeled files such as music or video, and the actual search is often limited to text tags. The explosive ...
ICDM
2009
IEEE
13 years 5 months ago
2009
IEEE
Multi-task learning refers to the learning problem of performing inference by jointly considering multiple related tasks. There have already been many research efforts on supervise...
ICML
2008
IEEE
14 years 8 months ago
2008
IEEE
We propose a new stopping condition for a Support Vector Machine (SVM) solver which precisely reflects the objective of the Leave-OneOut error computation. The stopping condition ...
DMIN
2006
13 years 9 months ago
2006
In this paper, we propose a post randomization technique to learn a Bayesian network (BN) from distributed heterogeneous data, in a privacy sensitive fashion. In this case, two or ...