On-line Prediction with Kernels and the Complexity Approximation Principle.
Alexander GammermanYuri KalnishkanVladimir VovkPublished in: UAI (2004)
Keyphrases
- approximation error
- prediction accuracy
- neural network
- worst case
- computational complexity
- prediction model
- error tolerance
- data sets
- free energy
- space complexity
- approximation algorithms
- np hard
- support vector
- kernel methods
- memory requirements
- closed form
- decision problems
- queueing networks
- relative error
- special case
- feature space
- approximation schemes