Rule prepending and post-pruning approach to incremental learning of decision lists

Murthy, KRK and Keerthi, SS and Murty, MN (2001) Rule prepending and post-pruning approach to incremental learning of decision lists. In: Pattern Recognition, 34 (8). pp. 1697-1699.

PDF sdarticle.pdf Restricted to Registered users only Download (77kB) | Request a copy

Abstract

A decision list [1], DL, is defined as a list of ordered pairs $\{(T_1,V_1), (T_2,V_2),... , (T_r,V_r)\}$. These pairs are called nodes and they are denoted as $N_1,N_2,...,N_r$, where $N_i=(T_i,V_i). N_r$ is called default node of DL. Each $T_i$ is a test whose outcome is either True or False, each $V_i$ is a class label, and $T_r$ is the constant function, True. DL defines a classification function as follows: for any input x, DL(x) is defined to be equal to $V_j$, where j is the least index such that $T_j(x)$ = True. We denote the index of node $N_k$ as Index $(N_k)$, i.e. k=Index $(N_k)$.

Item Type:	Journal Article
Publication:	Pattern Recognition
Publisher:	Elsevier
Additional Information:	Copyright of this article belongs to Elsevier.
Keywords:	Decision list;Incremental learning;Rule induction;CDL4;Decision list pruning
Department/Centre:	Division of Electrical Sciences > Computer Science & Automation
Date Deposited:	30 Mar 2007
Last Modified:	19 Sep 2010 04:36
URI:	http://eprints.iisc.ac.in/id/eprint/10488

Actions (login required)

View Item