Introduction to the special issue on COLT 2006
DNF are teachable in the average case
Unconditional lower bounds for learning intersections of halfspaces
A primal-dual perspective of online learning algorithms
Tracking the best hyperplane with a simple budget Perceptron
Logarithmic regret algorithms for online convex optimization
Competing with wild prediction rules
Active sampling for multiple output identification
Surrogate maximization/minimization algorithms and extensions
Classifying under computational resource constraints: anytime classification using probabilistic estimators
Extending boosting for large scale spoken language understanding