Clearwater, S., & Stern, E. (1991). A rule-learning program in high energy physics event classication. Comp Physics Comm, 67, 159-182.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....a 99.9 accuracy. This accuracy may be quite dicult for an induction algorithm to beat, though the simple rule presumably is unacceptable if a non trivial solution is sought. Skews of 10 2 are common in fraud detection and skews exceeding 10 6 have been reported in other applications (Clearwater Stern, 1991). Evaluation by classi cation accuracy also assumes equal error costs: c(Y; n) c(N; p) In the real world classi cations lead to actions, which have consequences. Actions can be as diverse as denying a credit charge, discarding a manufactured part, moving a control surface on an airplane, or ....

Clearwater, S., & Stern, E. (1991). A rule-learning program in high energy physics event classication. Comp Physics Comm, 67, 159-182.

.... 4) or (position = core) or (atom name ISA heteroatom) If all the LHS clauses are true of an example, then the rule matches the example and predicts that the example is a member of the class named in the righthand side (RHS) RL has been used to learn filtering criteria for high energy physics (Clearwater and Stern 1991), rules to predict carcinogenicity of chemicals (Lee, Buchanan, et al. 1996) and diagnostic rules for telephone line problems (Danyluk and Provost 1993) Empirical discovery. Empirical discovery is the process of discovering regularities or patterns from a set of observations (Buchanan 1983) ....

Clearwater, S. and Stern, E. 1991. A Rule-learning Program in High Energy Physics Event Classification. Computer Physics Communications 67: 159-182.

....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 (1994) only 0.2 (1 in 500) examples are positive. In the high energy physics learning problem reported by Clearwater and Stern (1991), only 1 example in a million is positive. The third critical feature is that examples are naturally grouped in batches. The examples drawn from the same image constitute a single batch. Whenever data is collected in batches, there is a possibility that the batches systematically differ from one ....

Clearwater, S., & Stern, E. (1991). A rule-learning program in high energy physics event classification.

....induction algorithm to beat, though the simple rule presumably is unacceptable if a non trivial solution is sought. Skews of 10 2 are common in fraud detection and Running head: Robust Classification for Imprecise Environments 5 skews exceeding 10 6 have been reported in other applications (Clearwater Stern, 1991). Evaluation by classification accuracy also tacitly assumes equal error costs : c(Y; n) c(N; p) In the real world this is rarely the case because classifications tacitly involve actions, which have consequences. Actions can be as diverse as cancelling a credit card account, moving a control ....

Clearwater, S., & Stern, E. (1991). A rule-learning program in high energy physics event classification. Comp Physics Comm, 67, 159--182.

....A simple rule, always classify as the maximum likelihood class, gives a 99.9 accuracy. Presumably this is not satisfactory if a non trivial solution is sought. Skews of 10 2 are common in fraud detection and skews greater than 10 6 have been reported in other classifier learning applications (Clearwater Stern 1991). Evaluation by classification accuracy also tacitly assumes equal error costs that a false positive error is equivalent to a false negative error. In the real world this is rarely the case, because classifications lead to actions which have consequences, sometimes grave. Rarely are mistakes ....

Clearwater, S., and Stern, E. 1991. A rule-learning program in high energy physics event classification.

....enough to be applicable in a wide variety of scenarios, and e ective enough to succeed. We are most familiar with RL, which has been applied to problems including identi cation of human developmental toxicity (Gomez, Lee, Mattison, 1993, 1994) trigger design for high energy physics systems (Clearwater Stern, 1991), sensitivity analysis in high energy physics (Clearwater Lee, 1993) building systems for diagnosing telecommunication networks (Danyluk Provost, 1993) inducing rules for biological macro molecule crystallization (Hennessy, Gopalakrishnan, Buchanan, 7 Rosenberg, Subramanian, 1994) ....

Clearwater, S., & Stern, E. (1991). A rule-learning program in high energy physics event classication. Comp Physics Comm, 67, 159-182.

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