scispace - formally typeset
Search or ask a question
Topic

Decision tree model

About: Decision tree model is a research topic. Over the lifetime, 2256 publications have been published within this topic receiving 38142 citations.


Papers
More filters
Proceedings ArticleDOI
30 Aug 2006
TL;DR: A GEP decision tree (GEPDT) system is presented, the system can construct a decision tree for classification without priori knowledge about the distribution of data and can solve a n-class problem in a single run, preliminary results show that the performance of GEP based decision tree is comparable to IDS.
Abstract: Gene expression programming (GEP) is a kind of genotype/phenotype based genetic algorithm. Its successful application in classification rules mining has gained wide interest in data mining and evolutionary computation fields. However, current GEP based classifiers represent classification rules in the form of expression tree, which is less meaningful and expressive than decision tree. What’s more, these systems adopt one-against-all learning strategy, i.e. to solve a n-class with n runs, each run solving a binary classification task. In this paper, a GEP decision tree(GEPDT) system is presented, the system can construct a decision tree for classification without priori knowledge about the distribution of data, at the same time, GEPDT can solve a n-class problem in a single run, preliminary results show that the performance of GEP based decision tree is comparable to ID3.

19 citations

Book ChapterDOI
19 Feb 1987
TL;DR: This paper investigates the problem of making existing spanning tree algorithms fault-resilient, and still overcome these difficulties, and introduces amortized message complexity as a tool for analyzing the message complexity.
Abstract: We study distributed algorithms for networks with undetectable fail-stop failures, assuming that all of them had occurred before the execution started (It was proved that distributed agreement cannot be reached when a node may fail during execution) Failures of this type are encountered, for example, during a recovery from a crash in the network We study the problems of leader election and spanning tree construction, that have been characterized as fundamental for this environment We point out that in presence of faults just duplicating messages in an existing algorithm does not suffice to make it resilient; actually, this redundancy gives rise to synchronization problems and also might increase the message complexity In this paper we investigate the problem of making existing spanning tree algorithms fault-resilient, and still overcome these difficulties Several lower bounds and optimal fault-resilient algorithms are presented for the first timeHowever, we believe that the main contribution of the paper is twofold: First, in designing the algorithms we use tools that thus argued to be rather general (for example, we extend the notion of token algorithms to multiple-token algorithms) In fact we are able to use them on several different algorithms, for several different families of networks Second, following the amortized computational complexity, we introduce amortized message complexity as a tool for analyzing the message complexity

19 citations

Proceedings ArticleDOI
11 Oct 2009
TL;DR: Numerical simulations and theoretical analysis show this new multi-stage decision tree improves the performance of traditional decision tree and decreases the computational complexity a lot compare with large margin learning based multi- stage decision tree.
Abstract: Motivated by overcoming the drawbacks of traditional decision tree and improving the efficiency of large margin learning based multi-stage decision tree when dealing with multi-class classification problems, this paper proposes a novel Multi-stage Decision Tree algorithm based on inter-class and inner class margin of SVM. This new algorithm is well designed for multi-class classification problem based on the maximum margin of SVM and the cohesion and coupling theory of clustering. Considering the multi-class classification problem as a clustering problem, this new algorithm attempts to convert the multi-class classification problem into a two-class classification problem such that the highest cohesion degree within classes while lowest coupling degree between classes, where the margin of SVM is considered as the measurement of the degree. Then for each two-class problem, this paper uses traditional C4.5 algorithm to generate each stage decision tree which splits a dataset into two subsets for the further induction. Recursively, the Multi-stage decision tree is obtained. Numerical simulations and theoretical analysis show this new multi-stage decision tree improves the performance of traditional decision tree and decreases the computational complexity a lot compare with large margin learning based multi-stage decision tree.

19 citations

Journal ArticleDOI
TL;DR: A branch-and-bound algorithm is given for constructing an optimal decision tree (sequential evaluation procedure) and the tree is optimal in minimizing the average number of variables which need to be examined.
Abstract: For monotonic Boolean functions, a branch-and-bound algorithm is given for constructing an optimal decision tree (sequential evaluation procedure). The tree is optimal in minimizing the average number of variables which need to be examined.

19 citations


Network Information
Related Topics (5)
Cluster analysis
146.5K papers, 2.9M citations
80% related
Artificial neural network
207K papers, 4.5M citations
78% related
Fuzzy logic
151.2K papers, 2.3M citations
77% related
The Internet
213.2K papers, 3.8M citations
77% related
Deep learning
79.8K papers, 2.1M citations
77% related
Performance
Metrics
No. of papers in the topic in previous years
YearPapers
202310
202224
2021101
2020163
2019158
2018121