NIPS
2004
13 years 5 months ago
2004
Classification algorithms typically induce population-wide models that are trained to perform well on average on expected future instances. We introduce a Bayesian framework for l...