## Linear Machine Decision Trees (1991)

Citations: | 36 - 1 self |

### BibTeX

@TECHREPORT{Utgoff91linearmachine,

author = {Paul E. Utgoff and Carla E. Brodley},

title = {Linear Machine Decision Trees},

institution = {},

year = {1991}

}

### Years of Citing Articles

### OpenURL

### Abstract

This article presents an algorithm for inducing multiclass decision trees with multivariate tests at internal decision nodes. Each test is constructed by training a linear machine and eliminating variables in a controlled manner. Empirical results demonstrate that the algorithm builds small accurate trees across a variety of tasks. 1 Introduction One of the fundamental research problems in machine learning is how to learn from examples. From a sequence or set of training examples, each labeled with its correct class name, a machine learns by forming or selecting a generalization of the training examples. This process, also known as supervised learning, is useful for real classification tasks, e.g. disease diagnosis, and for problem solving tasks in which control decisions depend on classification, e.g. rule applicability. The ability to generalize is fundamental to intelligence because it allows one to reason in accordance with predictions that are often correct. This article focuse...

### Citations

3918 | Pattern Classification and Scene Analysis - Duda, Hart - 1973 |

3903 | Classification and regression trees - Breiman, Friedman, et al. |

3351 | Induction of decision trees
- Quinlan
- 1986
(Show Context)
Citation Context ...8.3 23 97.59 84.88 Segment 7 17 1 5.8 5.8 98.86 94.25 3.2 Results for a Variety of Domains Table 4 show various performance measures for the LMDT algorithm on a variety of tasks. Pessimistic pruning (=-=Quinlan, 1987) is used to avoid o-=-verfitting. Each measure is an average for five runs. Column "Invars" shows the number of input variables. Column "Vars/LM" indicates the average number of variables per linear mac... |

308 |
Learning Efficient Classification Procedures and their ,ipplication to Chess Endgame
- Quinlan
(Show Context)
Citation Context ...s two tests and three leaves and is logically equivalent to that found by FRINGE. The root node has an LM based on variables a and b and the second node has an LM based on c, d and e. The chess task (=-=Quinlan, 1983-=-) demonstrates that LMDT finds small trees. The tree found by LMDT contains 7 linear machines, each with an average of 11.2 variables. The smallest tree produced by an ID3 variant (Utgoff, 1989) conta... |

161 | Incremental Induction of Decision Trees
- Utgoff
- 1989
(Show Context)
Citation Context ...task (Quinlan, 1983) demonstrates that LMDT finds small trees. The tree found by LMDT contains 7 linear machines, each with an average of 11.2 variables. The smallest tree produced by an ID3 variant (=-=Utgoff, 1989-=-) contains 62 univariate tests. A linear machine is more complex than a univariate test, so a strict comparison is unfair. The LED (Breiman, Friedman, Olshen & Stone, 1984) data set shows LMDT's perfo... |

150 | Learning Machines
- Nilsson
- 1965
(Show Context)
Citation Context ...ode based on a heuristic measure, such as information gain (Lewis, 1962), the LMDT algorithm trains a linear machine, which then serves as a multivariate test for the decision node. A linear machine (=-=Nilsson, 1965-=-; Duda & Hart, 1973) is a multiclass linear discriminant, which itself classifies the instance. The class name is the result of the linear machine test, hence there will be one branch for each possibl... |

118 | Connectionist expert systems - Gallant - 1988 |

111 | Learning by Being Told and Learning from Examples: An Experimental Comparison of the Two Methods of Knowledge Acquisition in the Context of Developing an Expert System for Soybean Disease Diagnosis, Int.J. of Policy Analysis and Info - Michalski, Chilausky - 1980 |

84 | An Experimental Comparison of Symbolic and Connectionist Learning Algorithms - Mooney, Shavlik, et al. - 1989 |

44 | A scaled conjugate gradient algorithm for fast supervised learning - Mller - 1993 |

34 | Linear function neurons: Structure and training - Hampson, Volper - 1986 |

21 | Improved decision trees: A generalized version of ID3 - Cheng, Fayyad, et al. - 1988 |

18 |
The characteristic selection problem in recognition systems
- Lewis
- 1962
(Show Context)
Citation Context ...a decision tree in the well known top-down manner, as indicated in Table 1. However, instead of selecting a univariate test for a decision node based on a heuristic measure, such as information gain (=-=Lewis, 1962-=-), the LMDT algorithm trains a linear machine, which then serves as a multivariate test for the decision node. A linear machine (Nilsson, 1965; Duda & Hart, 1973) is a multiclass linear discriminant, ... |

16 | nets and short paths: Optimising Neural Computation - Frean - 1990 |

11 | Pattern classification by iteratively determined linear and piecewise linear discriminant functions - DIJDA, FOSSUM - 1966 |