Learning by Experimentation: Acquiring and Refining Problem-Solving Heuristics

Mitchell, Tom M.; Utgoff, Paul E.; Banerji, Ranan B.

doi:10.1007/978-3-662-12405-5_6

Cited by 128 publications

(24 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In many applications such as mathematics and natural language processing, however, the instances are naturally represented as expressions or parse trees. For example, in symbolic integration, one might want to learn the set of expressions for which integration by parts is a good approach to use [Mitchell, Utgoff, & Banerji, 19831. In natural language processing, one might want to identify all (parsed) messages which contain a reference to a job opportunity or an automobile accident [Cardie, 19971. In many cases, such sets of expressions can be naturally represented as a union of "tree patterns".…”

Section: Introductionmentioning

confidence: 99%

Exact learning of tree patterns from queries and counterexamples

Amoth

Cull

Tadepalli

1998

Proceedings of the Eleventh Annual Conference on Computational Learning Theory

View full text Add to dashboard Cite

We consider learning tree patterns from queries. The instances are ordered and unordered trees with nodes labeled by constant identifiers. The concepts are tree patterns and unions of tree patterns (forests) where all the internal nodes are labeled with constants and the leaves are labeled with constants or variables. A tree pattern matches any tree with its variables replaced with constant subtrees. We show that ordered trees, in which the children are matched in a strict left-to-right order, are exactly learnable from equivalence queries, while ordered forests are learnable from equivalence and membership queries. Unordered trees are exactly leamable from superset queries, and unordered forests are learnable from super-set and equivalence queries. Negatively, we also show that each of the query types used is necessary for learning each concept class.

show abstract

Section: Introductionmentioning

confidence: 99%

Exact learning of tree patterns from queries and counterexamples

Amoth

Cull

Tadepalli

1998

Proceedings of the Eleventh Annual Conference on Computational Learning Theory

View full text Add to dashboard Cite

show abstract

“…Each solution found (to any of the previously solved tasks) is stored in the system's storage. Schmidhuber work extends earlier work on bias-optimal search algorithms done by Hutter [9] (for bias-learning see also [12]). Inherently, such search must deal with the trade-off between:…”

Section: Lifelong Reinforcement Learningmentioning

confidence: 88%

“…More details related to reinforcement learning can be found in the literature, for example [12], pages 367-387.…”

Section: Biological Learningmentioning

confidence: 99%

EVM: Lifelong reinforcement and self-learning

Nowostawski

2009

2009 International Multiconference on Computer Science and Information Technology

View full text Add to dashboard Cite

Open-ended systems and unknown dynamical environments present challenges to the traditional machine learning systems, and in many cases traditional methods are not applicable. Lifelong reinforcement learning is a special case of dynamic (process-oriented) reinforcement learning. Multi-task learning is a methodology that exploits similarities and patterns across multiple tasks. Both can be successfully used for open-ended systems and automated learning in unknown environments. Due to its unique characteristics, lifelong reinforcement presents both challenges and potential capabilities that go beyond traditional reinforcement learning methods. In this article, we present the basic notions of lifelong reinforcement learning, introduce the main methodologies, applications and challenges. We also introduce a new model of lifelong reinforcement based on the Evolvable Virtual Machine architecture (EVM).

show abstract

“…In the build time module, binary decision tree operators are automatically generated by a 'learning by experimentation' (Mitchell 1983) line simulator that can simulate the behaviour of the specific production line from an arhitrary initial status is indispensable for this module. The following tasks are performed in this module;…”

Section: Build Time Modulementioning

confidence: 99%

Dynamic scheduling system utilizing machine learning as a knowledge acquisition tool

Nakasuka

Yoshida

1992

International Journal of Production Research

View full text Add to dashboard Cite

Dynamic scheduling system utilizing machine learning as a knowledge acquisition tool SHINICHI NAKASUKAt and TAKETOSHI YOSHIDAtDynamicselectionofschedulingrulesduring real operations has beenrecognized as a promisingapproach to the schedulingof the production line. For this strategy to work effectively, sufficient knowledgeis required to enable prediction of which rule is the best to use under the current line status. In this paper, a new learning algorithm for acquiring such knowledge is proposed. In this algorithm, a binary decision tree is automatically generated using empirical data obtained by iterative production line simulations, and it decides in real time which rule to be used at decision points during the actual production operations. The configuration of the developed dynamic schedulingsystem and the learning algorithm are described in detail. Simulation results on its application to the dispatching problem are discussed with regard to its scheduling performance and learning capability.

show abstract

Learning by Experimentation: Acquiring and Refining Problem-Solving Heuristics

Cited by 128 publications

References 3 publications

Exact learning of tree patterns from queries and counterexamples

Exact learning of tree patterns from queries and counterexamples

EVM: Lifelong reinforcement and self-learning

Dynamic scheduling system utilizing machine learning as a knowledge acquisition tool

Contact Info

Product

Resources

About