P. Riddle, R. Segal and O. Etzioni. Representation design and brute-force induction in a Boeing manufacturing design. Applied Artificial Intelligence, 8:125-147, 1994.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....Wrobel, 1992] performs a complete search for clauses consistent with examples and user imposed constraints; however it uses a more restricted languages for constraints. Also, Minton s BFSF search strategy performs a complete search over the set of small first order formulae, and the BRUTE system [Riddle and ad Oren Etzioni, 1994] performs a similar search over the set of small conjunctive propositional rules. However, none of these earlier systems combine a complete search for all consistent hypotheses with an expressive constraint language. Grendel2 MD also appears to be unique in being able to construct hypotheses ....

Patricia Riddle and Richard Segal ad Oren Etzioni. Representation design and brute-force induction in a Boeing manufacturing domain. Applied Artificial Intelligence, 8:125--147, 1994.

....able to pac learn a concept and being able to perform data compression on a set of examples. One simple Occam algorithm is to enumerate the concepts in L in increasing order of complexity and output the rst hypothesis consistent with the examples. Although there are exceptions [ Minton, 1994; Riddle et al. 1994 ] such brute force learning systems are typically not practical for reasons of eciency. In the pac learning setting, enumerative learning methods usually satisfy all the requirements of pac learning except for the requirement of eciency. 2.2 Background knowledge These learning models are ....

P. Riddle, R. Segal, and O. Etzioni. Representation design and bruteforce induction in a Boeing manufacturing domain. Applied Arti cial Intelligence, 8:125{ 147, 1994.

....able to pac learn a concept and being able to perform data compression on a set of examples. One simple Occam algorithm is to enumerate the concepts in L in increasing order of complexity and output the first hypothesis consistent with the examples. Although there are exceptions [Minton, 1994, Riddle et al. 1994] such brute force learning systems are typically not practical for reasons of efficiency. In the pac learning setting, enumerative learning methods usually satisfy all the requirements of pac learning except for the requirement of efficiency. 2.2 Background knowledge These learning models are ....

P. Riddle, R. Segal, and O. Etzioni. Representation design and bruteforce induction in a Boeing manufacturing domain. Applied Artificial Intelligence, 8:125-- 147, 1994. 37

....pruning. He also provides clever methods for dynamic search space reordering that yield impressive speedups. The ITRULE algorithm (Smyth Goodman, 1992) conducts a gat rule space search with branch and bound pruning, using an information theory based measure of interestingness. The Brute programs (Riddle, Segal, Etzioni, 1994; Segal Etzioni, 1994) perform (highly optimized) gat rule space searches using Laplace corrected con dence and complexity bounds to de ne interestingness. Weiss, et al. Weiss, Galen, Tadepalli, 1990) describe a gat rule learning search program (PVM) and several pruning heuristics for ....

Riddle, P., Segal, R., & Etzioni, O. (1994). Representation design and brute-force induction in a boeing manufacturing domain. Applied Articial Intelligence, 8, 125-147.

....In short, one has to make a careful decision on which simplifications to the real world problem will lead to a model which still realistic and with which one can still work. 8 6 Develop new representations to allow experts to express their knowledge Example 8. Improving manufacturing processes (Riddle, Segal, and Etzioni, 1994) The Boeing company decided to use ML in order to improve some of its manufacturing processes. There are many problems to solve prior to applying the induction mechanism, and they are all more or less of a KA nature. These authors find five problems to address to begin with, and we shall point at ....

Riddle P., Segal R., Etzioni O. "Representation Design and Brute-Force Induction in a Boeing Manufacturing Domain," Applied Artificial Intelligence, special issue on applications of ML, 1994.

.... A different style of rule learning can be traced back to the search based data mining program MetaDENDRAL (Buchanan, Smith, White, Gritter, Feigenbaum, Lederberg, and Djerassi 1976) Buchanan and Feigenbaum 1978) Examples of MetaDENDRAL style rule learning include the Brute programs (Riddle, Segal, and Etzioni 1994; Segal and Etzioni 1994a) PVM (Weiss, Galen, and Tadepalli 1990) ITRULE (Smyth and Goodman 1992) the RL programs (Clearwater and Provost 1990; Provost and Buchanan 1995; Fawcett and Provost 1997) SE trees (Rymon 1993) and even Schlimmer s determination learning algorithm (Schlimmer 1993) ....

Riddle, P., R. Segal, and O. Etzioni (1994). Representation design and brute-force induction in a boeing manufacturing domain. Applied Artificial Intelligence 8, 125--147.

....with the data to be used in an application, and choose what works the best. This situation keeps the development and analysis of learning algorithms an open research issue. In spite of being frequently stated that exhaustive search for knowledge in databases is not practical, systems like Brute [RID94] have shown that this kind of approach avoids many of the pitfalls of using greedy techniques. Its empirical analysis has proved that it performs better than greedy algorithms in terms of accuracy, and that its cost is often reasonable. Based on such statements, exhaustive techniques could be ....

RIDDLE, P.; SEGAL, R.; ETZIONI, O. Representation design and brute-force induction in a Boeing manufacturing domain. Applied Artificial Intelligence, 8: 125-147, 1994. Available via anonymous FTP from ftp.cs.washington.edu, file /pub/ai/brute-aai94.ps.Z.

....A Mining Algorithm We treat the problem of mining contrast sets as a tree search problem. The root node is an empty contrastset, and we generate children of a node by specializing the set by adding one more term. We use a canonical ordering of attributes to avoid visiting the same node twice [9]. Children are formed by appending terms that follow all existing terms in a given ordering. For example, consider an artificial domain with two attributes, A 1 = fV 11 ; V 12 g and A 2 = fV 21 ; V 22 g, each with two possible values. Figure 2 shows the resulting search tree and enumerates every ....

P. Riddle, R. Segal, and O. Etzioni. Representation design and brute-force induction in a Boeing manufacturing domain. Applied Artificial Intelligence, 8:125--147, 1994.

.... Highly imbalanced training sets occur in applications where the classifier is to detect a rare but important event, such as fraudulent telephone calls (Fawcett Provost, 1997) unreliable telecommunications customers (Ezawa, Singh Norton, 1996) failures or delays in a manufacturing process (Riddle, Segal Etzioni, 1994), rare diagnoses such as the thyroid diseases in the UCI repository (Murphy Aha, 1994) or carcinogenicity of chemical compounds (Lee, Buchanan Aronis, this issue) Extremely imbalanced classes also arise in information retrieval and filtering tasks. In the domain studies by Lewis and Catlett ....

....insensitive to the class distribution in the training set. Extreme examples of this are algorithms that learn from positive examples only. A less extreme approach is to learn from positive and negative examples but to learn only rules that predict the positive class, as is done by BRUTE (Riddle et al. 1994). By measuring performance only of the positive predicting rules BRUTE is not influenced by the invariably high accuracy on the negative examples that are not covered by the positive predicting rules. Our SHRINK algorithm (Kubat et al. 1997) follows the same general principle find the rule ....

Riddle, P., Segal, R., & Etzioni, O. (1994). Representation design and brute-force induction in a Boeing manufacturing domain. Applied Artificial Intelligence, 8, 125--147.

.... for increasing productivity [243] for material procurement method selection [103] to accelerate rotogravure printing [126] for process optimization in electrochemical machining [130] to schedule printed circuit board assembly lines [383] to uncover flaws in a Boeing manufacturing process [407] and for quality control [185] For a recent review of the use of machine learning (decision trees and other techniques) in scheduling, see [14] ffl Medicine: Medical research and practice have long been important areas of application for decision tree techniques. Recent uses of automatic ....

P. Riddle, R. Segal, and O. Etzioni. Representation design and brute-force induction in a Boeing manufacturing domain. Applied Artificial Intelligence, 8(1):125-- 147, January-March 1994.

....patterns is called the lag of the rule. msdd finds the k strongest dependencies in a dataset by conducting a systematic search in the space of possible dependencies. Systematic search expands the children of search nodes in a manner that ensures that no node can ever be generated more than once [OGC95a, RSE94, Rym92, Sch93, Web96]. Because non redundant expansion is achieved without access to large, rapidly changing data structures such as lists of open and closed nodes, the search space can be divided into many computationally independent subsets, each of which may be processed in parallel. Distributed msdd (dmsdd) is an ....

Patricia Riddle, Richard Segal, and Oren Etzioni. Representation design and brute-force induction in a boeing manufacturing domain. Applied Artificial Intelligence, 8:125--147, 1994.

....are created from training examples. Nearest neighbour [CS93] is a classical approach to classify from numerical information. Again, to solve a classification problem using nearest neighbour approach, training examples have to be provided. Decision trees [Qui93, ADW94] and decision lists [RSE94] are classification methods working not only for numerical data but also for symbolic information. However, it is not possible to take relationships between objects into account. For example, whether a firm is interesting or not can not take into account information about this firm s subcompanies. ....

P. Riddle, R. Segal, and O. Etzoni. Representation design and brute-force induction in a Boeing manufacturing domain. Applied Artificial Intelligence, pages 125--147, 1994.

....C4.5 exhibit considerable variance. This can be used to advantage in getting different views of the data. 30.6 Related work Perhaps consistent with the fact that most decision tree induction algorithms offer no way of requesting greater sensitivity to one class, Segal, Etzioni, Riddle [21] find standard methods biased towards producing full coverage classifiers functions that seek to map any example to the appropriate class. As a result, the algorithms can overlook patterns that occur in small subsets of the data. Yet, in our application, even minute patterns represent valuable ....

Patricia Riddle, Richard Segal, and Oren Etzioni. Representation design and bruteforce induction in a boeing manufacturing domain. Applied Artificial Intelligence, 8:125--147, 1994.

.... [163] for increasing productivity [179] for material procurement method selection [73] to accelerate rotogravure printing [92] for process optimization in electro chemical machining [95] to schedule printed circuit board assembly lines [296] to uncover flaws in a Boeing manufacturing process [313] and for quality control [135] For a recent review of the use of machine learning (decision trees and other techniques) in scheduling, see [13] ffl Medicine: Medical research and practice have long been important areas of application for decision tree techniques. Recent uses of automatic ....

P. Riddle, R. Segal, and O. Etzioni. Representation design and brute-force induction in a Boeing manufacturing domain. Applied Artificial Intelligence, 8(1):125--147, January-March 1994.

....of its environment. A model of those dynamics is constructed based only on the agent s own past interactions with its environment. msdd s approach to expanding the search tree to avoid redundant generation of search nodes is similar to that of other algorithms (Rymon 1992) Schlimmer 1993) (Riddle, Segal, Etzioni 1994). msdd s search differs from those mentioned above in that it explores the space of rules containing both conjunctive lefthand sides and conjunctive right hand sides. Doing so allows msdd to find structure in the agent s interactions with its environment that could not be found by the ....

Riddle, P.; Segal, R.; and Etzioni, O. 1994. Representation design and brute-force induction in a boeing manufacturing domain. Applied Artificial Intelligence 8:125--147.

....collaboration with Riddle at Boeing, based on their experiences using the IND algorithm [2] on Boeing datasets. IND simulates several decision tree algorithms including CART and C4.5. The difficulty in using decision tree algorithms in these domains led to the development of Brute, as discussed in [6]. This paper also contains a thorough 346 Pat Riddle, Roman Fresnedo, David Newman description of the Brute algorithm and experiments comparing it to CART and C4.5. The major focus in developing Brute was to develop an algorithm which would discover a few good rules which may only cover a small ....

....work. The Brute algorithm itself is most similar to ITRULE. The evaluation function they used is more tuned to finding high coverage rules instead of high predictive rules as is the Brute evaluation function. A comparison between Brute, ITRULE and other rule extraction systems can be seen in [6]. Our current visualization work is a small portion of the field of scientific visualization. We have not delved into the more advanced techniques in this field. Currently the more mundane aspects of this field have been sufficient for our framework. Likewise the field of data engineering is vast ....

P.J. Riddle, R. Segal, and O. Etzioni. Representation design and brute-force induction in a Boeing manufacturing domain. Applied Artificial Intelligence, 8(1):125--147, 1994.

....Otherwise, than layered search should have performed much better than massive search since it is learning more accurate rules. 6 Related Work This paper has its foundations in several recent systems that use massive search [Smyth and Goodman, 1991, Webb, 1993, Murphy and Pazzani, 1994, Riddle et al. 1994, Segal and Etzioni, 1994, Agrawal et al. 1993] Most of these systems use Laplace error to evaluate rules and have noticed a decrease in performance due to additional search. One notable exception is Brute [Riddle et al. 1994] a system that applies massive search to data mining. Brute ....

....search [Smyth and Goodman, 1991, Webb, 1993, Murphy and Pazzani, 1994, Riddle et al. 1994, Segal and Etzioni, 1994, Agrawal et al. 1993] Most of these systems use Laplace error to evaluate rules and have noticed a decrease in performance due to additional search. One notable exception is Brute [Riddle et al. 1994], a system that applies massive search to data mining. Brute evaluated rules using rule accuracy and employed a minimum coverage requirement to avoid overfitting. On a Boeing manufacturing database, Brute found significantly better rules than those found by greedy algorithms. However, Brute used a ....

Patricia Riddle, Richard Segal, and Oren Etzioni. Representation design and brute-force induction in a Boeing manufacturing domain. Applied Artificial Intelligence, 8:125--147, 1994.

No context found.

Patricia Riddle, Richard Segal, and Oren Etzioni. Representation design and brute-force induction in a Boeing manufacturing domain. Applied Artificial Intelligence, 8:125--147, 1994. Available via anonymous FTP from /pub/ai at cs.washington.edu.

No context found.

P. Riddle, R. Segal and O. Etzioni. Representation design and brute-force induction in a Boeing manufacturing design. Applied Artificial Intelligence, 8:125-147, 1994.

No context found.

Riddle, P., Segal, R., Etzioni, O. Representation design and brute-force induction in a Boeing manufacturing design. Applied Artificial Intelligence 1994; 8:125-147. Schapire, R. E. A brief introduction to boosting. Proceedings of the Sixteenth International Joint Conference on Artificial Intelligence, 1999.

No context found.

Patricia Riddle, Richard Segal, and Oren Etzioni. Representation design and bruteforce induction in a boeing manufacturing domain. AppliedArtificial Intelligence, 8:125--147, 1994.

No context found.

P. Riddle, R. Segal, and O. Etzioni. Representation design and brute-force induction in a Boeing manufacturing domain. AppliedArti#cial Intelligence, 8#1#:125#147, January-March 1994.

No context found.

Riddle, P., Segal, R. and Etzioni, O., 1994. "Representation Design and BruteForce Induction in a Boeing Manufacturing Domain. Applied Articial Intelligence, 8:125-147.

No context found.

Patricia Riddle, Richard Segal, and Oren Etzioni. Representation design and brute-force induction in a boeing manufacturing domain. Applied Artificial Intelligence, 8:125--147, 1994.

No context found.

Patricia Riddle, Richard Segal, and Oren Etzioni. Representation design and bruteforce induction in a boeing manufacturing domain. Applied Artificial Intelligence, 8:125--147, 1994.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC