Murphy,P.,&Pazzani, M. (1991). ID2-of-3: Constructive induction of m-of-n concepts for discriminators in decision trees. Machine Learning: Proceedings of the Eighth International Workshop (pp. 183-187). Evanston, IL: Morgan Kaufmann.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....features. There are also many methods for the domain of binary and nominal features. The algorithms FRINGE and GREEDY3 (Pagallo Haussler 1989) create new Boolean features only by using logical operators to construct new features that are more adequate to describe hypotheses. ID2 of 3 (Murphy Pazzani 1991) creates M of N representations as new Boolean features. X of N (Zheng 1995) can be considered as an extension of M of N, that constructs new nominal features. The common main advantage of all constructive induction algorithms lies in the stronger expressive power of a concept found by these ....

Murphy, P. M., Pazzani, M. J. (1991). ID2-of-3: Constructive induction of M-of-N concepts for discriminators in decision trees, In Proceedings of the 8th International Machine Learning Workshop, Morgan Kaufmann.

....the overall accuracy is significantly reduced. In this thesis the learning algorithms used for meta learning are off the shelf algorithms and are the same as the base learning algorithms. More specialized metalevel attributes and algorithms can be devised. Learning algorithms that search Mof N (Murphy Pazzani, 1991) and other counting related concepts might be useful in locating effective combining rules. Constructive induction techniques (Matheus Rendell, 1989; Rendell, 1990) could also be beneficial in creating potentially relevant attributes. Moreover, learning algorithms that can incorporate weighted ....

Murphy, P. & Pazzani, M. (1991). ID2-of-3: constructive induction of M-of-N concepts for discriminators in decisions trees. Proc. Eigth Intl. work. Machine Learning (pp. 183--187).

....needs to represent an M of N concept, which is very difficult using DNF expressions. In recent years there have been several efforts on learning M of N concepts. Spackman (1998) developed the CRLS system that learns M of N rules by employing non equivalence symmetry bias and criteria tables. Murphy Pazzani (1991) developed the ID 2 of 3 system that incorporates M ofN tests in decision trees. Bloedorn Michalski (1991) developed AQ17 DCI a program that employs a variety of operators to construct new attributes. Fawcett and Utgoff (1991) used an attribute representation similar to the Michalski s (1983) ....

Murphy, P. M. & Pazzani, M. J. (1991). ID2-of-3: Constructive Induction of M-of-N Concepts for Discriminators in Decision Trees. In Proceedings of the 8th International Workshop on Machine Learning (pp. 183-187). Evanston, IL: Morgan Kaufmann.

....induction algorithms such as Fringe (Pagallo, 1990) Citre (Matheus Rendell, 1989) CI (Zheng, 1992) LFC (Ragavan Rendell, 1993) and CAT (Zheng, 1998) only construct new binary attributes by using logical operators such as conjunction, negation, and disjunction. On the other hand, ID2 of 3 (Murphy Pazzani, 1991) creates at least M of N attributes. An M of N representation consists of a set of conditions (attribute value pairs) and a value M . For a given instance, the value of an at least M of N representation is true if at least M of its conditions are true of the instance; it is false, otherwise. ....

....for decision tree learning. Tests at decision nodes of a decision tree are either primitive attributes or new nominal attributes in the form of X of N representations. During the generation of a tree, the construction of X of N attributes occurs. 3.1. Building decision trees Like ID2 of 3 (Murphy Pazzani, 1991), XofN consists of a single process while other constructive induction algorithms such as Fringe (Pagallo, 1990) and AQ17 hci (Wnek Michalski, 1994) interleave two processes, namely selective induction and new attribute construction. As shown in Table 1, XofN recursively builds a decision tree ....

[Article contains additional citation context not shown here]

Murphy, P., & Pazzani, M. (1991). ID2-of-3: Constructive induction of M-of-N concepts for discriminators in decision trees. Proceedings of the Eighth International Workshop on Machine Learning (pp. 183--187). San Mateo, CA: Morgan Kaufmann.

....is a special case of the M ofN concept. There have been several efforts concerned with learning Mof N concepts. For example, the system CRLS learns Mof N rules by employing non equivalence symmetry bias and criteria tables (Spackman, 1988) ID 2 of 3 incorporates M of N tests in decision trees (Murphy Pazzani, 1991), AQ17 DCI (Bloedorn Michalski, 1991) employs a variety of operators to construct new attributes, NEITHER MofN (Baffes Mooney, 1993) is able to refine M of N rules by increasing or decreasing either of M or N. The idea of counting attributes and counting arguments rule is related to ....

Murphy, P. M. and Pazzani, M. J., "ID2-of-3: Constructive Induction of M-of-N Concepts for Discriminators in Decision Trees," Proceedings of the 8th International Workshop on Machine Learning, Evanston, Ill., pp. 183-187, 1991.

....evaluate the algorithm. 3.1 TREPAN: An Algorithm for Extracting Decision Trees Our algorithm, called Trepan, 2 is shown in Table 1. Trepan is similar to conventional decisiontree induction algorithms, such as CART (Breiman et al. 1984) ID3 (Quinlan, 1986) C4.5 (Quinlan, 1993) and ID2 of 3 (Murphy Pazzani, 1991), which learn directly from a training set. These algorithms build decision trees by recursively partitioning the input space. Each internal node in such a tree represents a partition of some part of the input space, and each leaf represents a predicted class. Our method is substantially different ....

....than two values, we consider binary splits based on each allowable value of the feature (e.g. color=red , color=blue , The selected binary split serves as seed for the search process. The search uses information gain as its heuristic evaluation function, and uses the two following operators (Murphy Pazzani, 1991): ffl m of n 1 : Add a new value to the set, and hold the threshold constant. For example, 2 of fa; bg = 2 of fa; b; cg. ffl m 1 of n 1: Add a new value to the set, and increment the threshold. For example, 2 of fa; b; cg = 3 of fa; b; c; dg. Trepan s search for M of N splits differs ....

Murphy, P. M. & Pazzani, M. J. (1991). ID2-of-3: Constructive induction of M-of-N concepts for discriminators in decision trees. In Proceedings of the Eighth International Machine Learning Workshop, (pp. 183--187), Evanston, IL. Morgan Kaufmann.

....literals are satisfied. For example, suppose we have three Boolean features, x 1 , x 2 , and x 3 ; the m of n expression 2 of fx 1 ; x 2 ; x 3 g is logically equivalent to (x 1 :x 2 ) x 1 x 3 ) x 2 x 3 ) The advantage of using m of n tests is that they often result in more concise trees. Murphy and Pazzani (1991) introduced the idea of using m of n expressions as splitting criteria in decision trees, and Trepan s method for constructing such tests is patterned after their ID2 of 3 algorithm. Like the ID2 of 3 algorithm, Trepan uses a heuristic search process to construct its m of n splitting tests. In ....

....color=blue , For continuous features we consider binary tests on thresholds, as is done by C4.5. The selected binary test serves as a seed for the m of n search process. This search uses the information gain measure as its heuristic evaluation function, and uses the following two operators (Murphy Pazzani, 1991): ffl m of n 1 : Add a new literal to the set, and hold the threshold constant. For example, 2 of fx 1 ; x 2 g = 2 of fx 1 ; x 2 ; x 3 g. ffl m 1 of n 1: Add a new literal to the set, and increment the threshold. For example, 2 of fx 1 ; x 2 ; x 3 g = 3 of fx 1 ; x 2 ; x 3 ; x 4 g. Trepan ....

[Article contains additional citation context not shown here]

Murphy, P. M. & Pazzani, M. J. (1991). ID2-of-3: Constructive induction of M-of-N concepts for discriminators in decision trees. In Proceedings of the Eighth International Machine Learning Workshop, (pp.

....of the tree to the network for those instances. Split Types. The role of internal nodes in a decision tree is to partition the input space in order to increase the separation of instances of different classes. In C4.5, each of these splits is based on a single feature. Our algorithm, like Murphy and Pazzani s (1991) ID2 of 3 algorithm, forms trees that use m of n expressions for its splits. An m of n expression is a Boolean expression that is specified by an integer threshold, m, and a set of n Boolean conditions. An m of n expression is satisfied when at least m of its n conditions are satisfied. For ....

..... For continuous features, we consider binary splits on thresholds, in the same manner as C4.5. The selected binary split serves as a seed for the m of n search process. This greedy search uses the gain ratio measure as its heuristic evaluation function, and uses the following two operators (Murphy Pazzani, 1991): ffl m of n 1 : Add a new value to the set, and hold the threshold constant. For example, 2 of fa; bg = 2 of fa; b; cg. ffl m 1 of n 1: Add a new value to the set, and increment the threshold. For example, 2 of fa; b; cg = 3 of fa; b; c; dg. Unlike ID2 of 3, Trepan constrains m of n ....

[Article contains additional citation context not shown here]

Murphy, P. M. & Pazzani, M. J. (1991). ID2-of-3: Constructive induction of M-of-N concepts for discriminators in decision trees. In Proc. of the 8th International Machine Learning Workshop, (pp. 183--187), Evanston, IL. Morgan Kaufmann.

....may be expressed as the logical conjunction of N conditions that should be met. The most accurate theory, though, is one in which any M of these N conditions holds. Expressing this more accurate theory in the DNF representation used to describe the initial theory would be cumbersome and unwieldy (Murphy Pazzani, 1991). Furthermore, arriving at the final theory using the refinement operators most suitable for DNF (drop condition, add condition, modify condition) would be a cumbersome task. But when an M of N representation is adopted, the refinement simply involves empirically finding the appropriate M, and the ....

Murphy, P., & Pazzani, M. (1991). ID2-of-3: Constructive induction of M-of-N concepts for discriminators in decision trees. In Proceedings of the 1991 International Conference on Machine Learning.

....that best splits the data, according to the information based heuristic entropy function from ID3. The chosen feature value, F , becomes the 1 of (F ) initial hypothesis that operators are then applied to. For a description and comparison of other approaches for constructing m of n hypotheses see (Murphy Pazzani, 1991). Table 1. M of N Constructive Induction Algorithm function generate m of n(Examples) Best Concept : initial concept(Examples) Best Cost : eval concept(Examples,Best Concept) repeat Concept : Best Concept; Cost : Best Cost; for all Op Instantiation(Concept) Temp Concept : Op ....

Murphy, P. M & Pazzani, M. J. (1991). ID2-of-3: Constructive induction of M-of-N concepts for discriminators in decision trees (Technical Report 91-37). Irvine: University of California, Department of Information and Computer Science.

No context found.

Murphy,P.,&Pazzani, M. (1991). ID2-of-3: Constructive induction of m-of-n concepts for discriminators in decision trees. Machine Learning: Proceedings of the Eighth International Workshop (pp. 183-187). Evanston, IL: Morgan Kaufmann.

No context found.

Murphy, P. M., Pazzani, M. J. (1991). ID2-of-3: Constructive induction of M-of-N concepts for discriminators in decision trees, In Proceedings of the 8th International Machine Learning Workshop, Morgan Kaufmann.

No context found.

Murphy, P. and Pazzani, M. (1991). ID2-of-3: constructive induction of m-of-n concepts for discriminators in decision trees. Proceedings of the Eighth International Workshop on Machine Learning (ML91). San Mateo, CA: Morgan Kaufmann.

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