Learning decision trees from random examples needed for learning
Information and Computation
Mistake bounds and logarithmic linear-threshold learning algorithms
Mistake bounds and logarithmic linear-threshold learning algorithms
Introduction to algorithms
Equivalence queries and approximate fingerprints
COLT '89 Proceedings of the second annual workshop on Computational learning theory
Learning boolean functions in an infinite attribute space
STOC '90 Proceedings of the twenty-second annual ACM symposium on Theory of computing
On-line learning with an oblivious environment and the power of randomization
COLT '91 Proceedings of the fourth annual workshop on Computational learning theory
Learning Automata from Ordered Examples
Machine Learning - Connectionist approaches to language learning
On-line learning of rectangles
COLT '92 Proceedings of the fifth annual workshop on Computational learning theory
Rank-r decision trees are a subclass of r-decision lists
Information Processing Letters
Learning binary relations and total orders
SIAM Journal on Computing
STOC '93 Proceedings of the twenty-fifth annual ACM symposium on Theory of computing
The weighted majority algorithm
Information and Computation
The Power of Self-Directed Learning
Machine Learning
Separating Distribution-Free and Mistake-Bound Learning Models over the Boolean Domain
SIAM Journal on Computing
Learning classifier systems: a complete introduction, review, and roadmap
Journal of Artificial Evolution and Applications
Adaptive Versus Nonadaptive Attribute-Efficient Learning
Machine Learning
Exploring Learnability between Exact and PAC
COLT '02 Proceedings of the 15th Annual Conference on Computational Learning Theory
Writing the BoK: designing for the networked learning environment of college students
DUX '05 Proceedings of the 2005 conference on Designing for User eXperience
Learning classifier systems: a complete introduction, review, and roadmap
Journal of Artificial Evolution and Applications
Web-based multi-agent system architecture in a dynamic environment
International Journal of Knowledge-based and Intelligent Engineering Systems
| Hi-index | 0.00 |
We present an off-line variant of the mistake-bound model oflearning. This is an intermediate model between the on-line learning model (Littlestone, 1988, Littlestone, 1989) and theself-directed learning model (Goldman, Rivest& Schapire, 1993, Goldman & Sloan, 1994). Just like in theother two models, a learner in the off-line model has to learn anunknown concept from a sequence of elements of the instance space onwhich it makes “guess and test” trials. In all models,the aim of the learner is to make as few mistakes as possible. Thedifference between the models is that, while in the on-line modelonly the set of possible elements is known, in theoff-line model the sequence of elements (i.e., theidentity of the elements as well as the order in which they are to bepresented) is known to the learner in advance. On the other hand, thelearner is weaker than the self-directed learner, which is allowed tochoose adaptively the sequence of elements presented to him.We study some of the fundamental properties of the off-line model. In particular, we compare the number ofmistakes made by the off-line learner on certain concept classes tothose made by the on-line and self-directed learners. We give boundson the possible gaps between the various models and show examplesthat prove that our bounds are tight.Another contribution of this paper is the extension of thecombinatorial tool of labeled trees to a unified approach thatcaptures the various mistake bound measures of all the modelsdiscussed. We believe that this tool will prove to be useful forfurther study of models of incremental learning.