#### DMCA

## Learning an Optimally Accurate Representation System (1994)

### Cached

### Download Links

- [www.cs.toronto.edu]
- [www.cs.toronto.edu]
- DBLP

### Other Repositories/Bibliography

Venue: | In ECAI Workshop on Theoretical Foundations of Knowledge Representation and Reasoning |

Citations: | 2 - 1 self |

### Citations

2335 | Random Graphs - Bollobás - 2001 |

1063 |
Negation as Failure
- CLARK
- 1978
(Show Context)
Citation Context ...] for the EBL systems is classification (i.e., determining whether a given element is, or is not, a member of some target class) rather than general derivation; and (ii) each uses negation-as-failure =-=[Cla78]-=- (a hardwired form of non-monotonicity) to classify negatively any sample that cannot be proved to be in the class. By contrast, our work can accommodate general queries, and deals with general defaul... |

959 | A measure of asymptotic efficiency for tests of a hypothesis based on the sum of observations - Chernoff - 1952 |

927 | Learning logical definitions from relations
- Quinlan
- 1990
(Show Context)
Citation Context ...ptions based on bit vectors or simple hierarchies, our work deals in the context of propositions; here too there is a history of results, dating back (at least) to Shapiro [Sha83], and including foil =-=[Qui90]-=- and the body of work on inductive logic programming [MB88]. While much of that research deals with monotonic (usually propositional or first order logic) theories and discusses ways of extendingssuch... |

907 | On the logic of theory change: Partial meet contraction and revision functions - Alchourròn, Gärdenfors, et al. - 1985 |

723 | Knowledge in Flux: Modeling the Dynamics of Epistemic States - Gardenförs - 1988 |

567 | An analysis of time-dependent planning - Dean, Boddy - 1988 |

529 |
Algorithmic Program DeBugging
- Shapiro
- 1983
(Show Context)
Citation Context ...f these systems learn descriptions based on bit vectors or simple hierarchies, our work deals in the context of propositions; here too there is a history of results, dating back (at least) to Shapiro =-=[Sha83]-=-, and including foil [Qui90] and the body of work on inductive logic programming [MB88]. While much of that research deals with monotonic (usually propositional or first order logic) theories and disc... |

406 | Connectionist learning procedures - Hinton - 1989 |

252 | Quantifying inductive bias: AI learning algorithms and valiant’s learning framework - Haussler - 1988 |

237 | Solving time-dependent planning problems - Boddy, Dean - 1989 |

223 | Preferred subtheories: An extended logical framework for default reasoning - Brewka - 1989 |

177 |
KRYPTON: a functional approach to knowledge representation
- J, Levesque
- 1983
(Show Context)
Citation Context ...ation process. 2 Framework This section first provides the general framework for our analysis, then describes the class of representation systems we will use. 2.1 General Analytic Framework Following =-=[Lev84]-=- and [DP91], we view a representation system R as a function that maps each query to its proposed answer; hence, R : Q 7! A, where Q is a (possibly infinite) set of queries, and A is the set of possib... |

145 |
Theorist: a logical reasoning system for default and diagnosis. The Knowledge Frontier: Essays in the Representation of Knowledge
- Poole, Goebel, et al.
- 1987
(Show Context)
Citation Context ... Theorist-Style Representation Systems While much of our analysis applies to representation systems in general, this paper focuses one particular form: stratified Theorist-style representation system =-=[PGA86]-=- [Bre89, vA90]. Here, each R i can be expressed as a set of factual information, a set of allowed hypotheses (each a simple type of default) and an 5 We assume Q is at most countably infinite to simpl... |

139 |
Changing the rules: a comprehensive approach to theory refinement
- Ourston, Mooney
- 1990
(Show Context)
Citation Context ...ations, for converting one representation system into another --- for instance eliminating some inappropriate sets of hypotheses [Coh90, Won91], or modifying the antecedents of individual rules (cf., =-=[OM90]-=-), etc. Each of these approaches can be viewed as using a set of transformations to navigate around a space of interrelated representation systems. We can then consider the same objective described ab... |

132 | Two theses of knowledge representation: Language restrictions, taxonomic classification, and the utility of representation services
- Doyle, Patil
- 1991
(Show Context)
Citation Context ...s. 2 Framework This section first provides the general framework for our analysis, then describes the class of representation systems we will use. 2.1 General Analytic Framework Following [Lev84] and =-=[DP91]-=-, we view a representation system R as a function that maps each query to its proposed answer; hence, R : Q 7! A, where Q is a (possibly infinite) set of queries, and A is the set of possible answers.... |

121 | Knowledge compilation using Horn approximations - Selman, Kautz - 1991 |

98 |
Nonmonotonic reasoning
- Reiter
- 1987
(Show Context)
Citation Context ...nting this initial information with some new hypothesis (or conjecture or default) that is plausible but not necessarily true; each particular collection of facts and hypotheses is a "default the=-=ory" [Rei87]-=-. Unfortunately, there can often be more than one such hypothesis, and these hypotheses (and hence the conclusions they respectively entail) may not be compatible; consider for example the Nixon diamo... |

95 | A study of explanation-based methods for inductive learning - Flann, Dietterich - 1989 |

64 | Optimal problem-solving search: All-or-none solutions - Simon, Kadane - 1975 |

48 | A statistical approach to solving the EBL utility problem
- Greiner, Jurisica
- 1992
(Show Context)
Citation Context ...a [ q ] holds for each hypothesis h i . There can, in some situations, be more efficient ways of estimating these values, for example, by using some Horn approximation to F [ fh i g; see [Gre92a] and =-=[GJ92]-=-. We can also simplify the computation if the h j hypotheses are not independent; e.g., if each corresponds to a set of sub-hypotheses. 4.4 Alternative Performance Measures and Transformations We have... |

46 |
Machine invention of first order predicates by inverting resolution
- Muggleton, Buntine
- 1988
(Show Context)
Citation Context ...deals in the context of propositions; here too there is a history of results, dating back (at least) to Shapiro [Sha83], and including foil [Qui90] and the body of work on inductive logic programming =-=[MB88]-=-. While much of that research deals with monotonic (usually propositional or first order logic) theories and discusses ways of extendingssuch theories, producing new theories that can return additiona... |

45 | The reference class - Kyburg - 1983 |

37 |
Generalising prioritization
- Grosof
- 1991
(Show Context)
Citation Context ...ve assumed that every ordering of hypotheses is meaningful. In some contexts, there may already be a meaningful partial ordering of the hypotheses, perhaps based on specificity or some other criteria =-=[Gro91]-=-. Here, we can still use OptAcc to complete the partial ordering, by determining the relative priorities of the initially incomparable elements. In some situations, we may be unable to answer certain ... |

29 | Hierarchical knowledge bases and efficient disjunctive reasoning - Borgida, Etherington - 1989 |

28 | Controlling backward inference - Smith - 1989 |

26 | Learning from textbook knowledge: A case study - Cohen - 1990 |

25 |
Default reasoning in semantic networks: a formalization of recognition and inheritance
- Shastri
- 1989
(Show Context)
Citation Context ...lso seeks good weakenings. Its goal however is an efficientsencoding; by contrast, we are seeking an accurate representation. Finally, the motivation underlying our work is similar to the research in =-=[Sha89]-=- and elsewhere, which also uses probabilistic information to identify the best default theory. Our research differs by using statistical sampling techniques to obtain estimates of the required distrib... |

24 |
Models of learning systems
- Buchanan, Mitchell, et al.
- 1978
(Show Context)
Citation Context ...anch of explanation-based learning (EBL) research seeks the appropriate "specialization" (read "weakenings") of a given theory [FD89, OM90, Paz88, Coh92]; however, (i) the underlyi=-=ng performance task [BMSJ78]-=- for the EBL systems is classification (i.e., determining whether a given element is, or is not, a member of some target class) rather than general derivation; and (ii) each uses negation-as-failure [... |

24 | A declarative approach to bias in concept learning - Russell, Grosof - 1987 |

21 | Abductive explanation-based learning: a solution to the multiple inconsistent explanation problem - Cohen - 1992 |

18 | Tractable approximate deduction using limited vocabularies - Dalal, Etherington - 1992 |

17 | Learning Useful Horn Approximations - Greiner, Schuurmans - 1992 |

16 | Probabilistic hill climbing : Theory and applications
- Greiner
- 1992
(Show Context)
Citation Context ...ks in a "batched incremental" mode, as it iteratively uses a set of samples to decide whether to climb to a new theory, or to terminate. There is also a strictly-incremental variant of this =-=algorithm [Gre92b]-=-, which observes samples one-by-one, and decides after each individual sample, whether to climb, terminate, or simply draw an additional sample; hence this variant can, in some situations, climb to be... |

13 | Measuring and improving the effectiveness of representations
- Greiner, Elkan
- 1991
(Show Context)
Citation Context ...he user to prefer, for example, a performance system that returns IDK in complex situations, rather than spend a long time returning the correct answer; or even allow it to be wrong in some instances =-=[GE91]. Of cours-=-e, the OptAcc-variant may have to consider other transformations, besides the simple "reordering the hypotheses" one discussed above. For example, if being wrong was much worse than being si... |

10 | Probably approximately optimal derivation strategies - Greiner, Orponen - 1991 |

9 | and Pekka Orponen. Probably approximately optimal derivation strategies - Greiner - 1991 |

9 | The Need for Bias in Learning Generalizations - Mitchell - 1991 |

9 | D.: Probabilistic hill-climbing
- Cohen, Greiner, et al.
- 1994
(Show Context)
Citation Context ...ally intractable. However, if this process is decidable (e.g., if we are dealing with propositional theories), then OptAcc will terminate with probability 1, as the space of strategies is finite; see =-=[CG91]-=-. Finally, each iteration in the OptAcc algorithm is polytime if the F [ fh i g j= ? q k computation is polytime; e.g., if we are dealing with propositional Horn theories or propositional 2-CNF, etc. ... |

8 | Computing reference classes - Loui - 1988 |

7 | On the sample complexity of finding good search strategies - Orponen, Greiner - 1990 |

6 | Curing Anomalous Extensions - Morris - 1987 |

5 |
Learning near-optimal horn approximations
- Greiner
- 1992
(Show Context)
Citation Context ... g j= ? q=O qa [ q ] holds for each hypothesis h i . There can, in some situations, be more efficient ways of estimating these values, for example, by using some Horn approximation to F [ fh i g; see =-=[Gre92a]-=- and [GJ92]. We can also simplify the computation if the h j hypotheses are not independent; e.g., if each corresponds to a set of sub-hypotheses. 4.4 Alternative Performance Measures and Transformati... |

5 | Selecting the best explanation in explanation-based learning - Pazzani - 1988 |

4 | Nested default reasoning with priority levels - Arragon - 1990 |

2 |
Evaluating answers to questions
- Vormittag
- 1991
(Show Context)
Citation Context ...nswer. As another situation, we may be able to rank responses in terms of their precision: e.g., knowing that the cost of watch 7 is $3;000 is more precise than knowing only that watch 7 is expensive =-=[Vor91].) We have-=- also assumed that all queries are equally important; i.e., a wrong answer to any query "costs" us the same 0, whether we are asking for the location of a salt-shaker, or of the tiger curren... |

2 | Improving the accuracy of a representational system - Wong - 1991 |

2 |
Producing more accurate representational systems
- Greiner
- 1992
(Show Context)
Citation Context ... considers a more general framework. 4 N.b., we assume that R will return a single answer. If there are several compatible binding lists, then R will select and return one of them; see extended paper =-=[GS92]-=-. is the hypothesis set, and and \Upsilon A = hh 1 ; h 2 i is the hypothesis ordering. 5 To explain how RA would process a query, imagine we want to know the color of Zelda --- i.e., we want to find a... |

1 |
The complexity of computing optimally-accurate default theories
- Greiner
- 1993
(Show Context)
Citation Context ... NP-complete. This holds even for the simplistic situation we have been considering, where every derivation requires exactly one hypothesis, every ordering of hypotheses is allowed, and so forth; see =-=[Gre93]-=-. 3 The OptAcc Algorithm This section presents a learning system, OptAcc, that side-steps the two problems mentioned above. OptAcc copes with the problem of an unknown query distribution by using a se... |