Explanation-based learning: a survey of programs and perspectives

Ellman, Thomas

doi:10.1145/66443.66445

Cited by 114 publications

(37 citation statements)

References 53 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the context of machine learning, our approach is an instance of Explanation-Based Learning (EBL) [9]. EBL refers to learning from a single example using an explanation of that example.…”

Section: Related Workmentioning

confidence: 99%

Generating compiler optimizations from proofs

2010

View full text Add to dashboard Cite

We present an automated technique for generating compiler optimizations from examples of concrete programs before and after improvements have been made to them. The key technical insight of our technique is that a proof of equivalence between the original and transformed concrete programs informs us which aspects of the programs are important and which can be discarded. Our technique therefore uses these proofs, which can be produced by translation validation or a proof-carrying compiler, as a guide to generalize the original and transformed programs into broadly applicable optimization rules.We present a category-theoretic formalization of our proof generalization technique. This abstraction makes our technique applicable to logics besides our own. In particular, we demonstrate how our technique can also be used to learn query optimizations for relational databases or to aid programmers in debugging type errors.Finally, we show experimentally that our technique enables programmers to train a compiler with application-specific optimizations by providing concrete examples of original programs and the desired transformed programs. We also show how it enables a compiler to learn efficient-to-run optimizations from expensive-to-run super-optimizers.

show abstract

“…In the context of machine learning, our approach is an instance of Explanation-Based Learning (EBL) [9]. EBL refers to learning from a single example using an explanation of that example.…”

Section: Related Workmentioning

confidence: 99%

Generating compiler optimizations from proofs

2010

View full text Add to dashboard Cite

show abstract

“…Indeed, this is akin to explanation-based learning (EBL) [19], [20], where subsequent to a successful problem solving session a proof is constructed that explains the success. The proof is then generalized to a description of states which can be solved in the same way.…”

Section: Relational Navigation Policiesmentioning

confidence: 99%

Learning Relational Navigation Policies

Cocora

Kersting

Plagemann

et al. 2006

2006 IEEE/RSJ International Conference on Intelligent Robots and Systems

View full text Add to dashboard Cite

Abstract-Navigation is one of the fundamental tasks for a mobile robot. The majority of path planning approaches has been designed to entirely solve the given problem from scratch given the current and goal configurations of the robot. Although these approaches yield highly efficient plans, the computed policies typically do not transfer to other, similar tasks. We propose to learn relational decision trees as abstract navigation strategies from example paths. Relational abstraction has several interesting and important properties. First, it allows a mobile robot to generalize navigation plans from specific examples provided by users or exploration. Second, the navigation policy learned in one environment can be transferred to unknown environments. In several experiments with real robots in a real environment and in simulated runs, we demonstrate the usefulness of our approach.

show abstract

“…In systems designed to support the teaching of logic a flexible facility for giving advice should know how to construct good proofs in the deductive system underlying the course. In systems where proofs are the data to be manipulated to fit other tasks such as explanation-based generalization [10] or the extraction of programs [5], the structure of proofs becomes of primary importance which, unfortunately, in practice prohibits the use of state-of-the-art theorem proving technology in such applications. Further evidence that proof presentation is a hard and important problem is that mathematicians spend a large percentage of their time analyzing and reformulating proofs.…”

Section: Introductionmentioning

confidence: 99%

Presenting intuitive deductions via symmetric simplification

Pfenning

1990

10th International Conference on Automated Deduction

View full text Add to dashboard Cite

In automated deduction systems that are intended for human use, the presentation of a proof is no less important than its discovery. For most of today's automated theorem proving systems, this requires a non-trivial translation procedure to extract human-oriented deductions from machine-oriented proofs. Previously known translation procedures, though complete, tend to produce unintuitive deductions. One of the major flaws in such procedures is that too often the rule of indirect proof is used where the introduction of a lemma would result in a shorter and more intuitive proof.We present an algorithm, symmetric simplification, for discovering useful lemmas in deductions of theorems in first-and higher-order logic. This algorithm, which has been implemented in the TPS system, has the feature that resulting deductions may no longer have the weak subformula property. It is currently limited, however, in that it only generates lemmas of the form C ∨ ¬C , where C and C have the same negation normal form.

show abstract

Explanation-based learning: a survey of programs and perspectives

Cited by 114 publications

References 53 publications

Generating compiler optimizations from proofs

Generating compiler optimizations from proofs

Learning Relational Navigation Policies

Presenting intuitive deductions via symmetric simplification

Contact Info

Product

Resources

About