990 search results - page 112 / 198 » Learning with non-positive kernels |
ICML
2008
IEEE
15 years 10 months ago
2008
IEEE
We show how nonlinear embedding algorithms popular for use with shallow semisupervised learning techniques such as kernel methods can be applied to deep multilayer architectures, ...
154
click to vote
VLSISP
2011
14 years 4 months ago
2011
Machine-learning algorithms are employed in a wide variety of applications to extract useful information from data sets, and many are known to suffer from superlinear increases in ...
127
click to vote
EWRL
2008
14 years 11 months ago
2008
A wide variety of function approximation schemes have been applied to reinforcement learning. However, Bayesian filtering approaches, which have been shown efficient in other field...
111
click to vote
EMNLP
2011
13 years 9 months ago
2011
This paper describes a novel approach to the semantic relation detection problem. Instead of relying only on the training instances for a new relation, we leverage the knowledge l...
ICML
2004
IEEE
15 years 3 months ago
2004
IEEE
We introduce a new perceptron-based discriminative learning algorithm for labeling structured data such as sequences, trees, and graphs. Since it is fully kernelized and uses poin...