Adaptive Directed Acyclic Graphs for Multiclass Classification

Authors:
Boonserm Kijsirikul;Nitiwut Ussivakul;Surapant Meknavin
Affiliations:
-;-;-
Venue:
PRICAI '02 Proceedings of the 7th Pacific Rim International Conference on Artificial Intelligence: Trends in Artificial Intelligence
Year:
2002

Citing 2
Cited 7

A Tutorial on Support Vector Machines for Pattern Recognition

Data Mining and Knowledge Discovery
Solving multiclass learning problems via error-correcting output codes

Journal of Artificial Intelligence Research

Erratum: Constructing Multiclass Learners from Binary Learners: A Simple Black-Box Analysis of the Generalization Errors

ALT '08 Proceedings of the 19th international conference on Algorithmic Learning Theory
CombNET-III with Nonlinear Gating Network and Its Application in Large-Scale Classification Problems

IEICE - Transactions on Information and Systems
Evaluation Measures for Multi-class Subgroup Discovery

ECML PKDD '09 Proceedings of the European Conference on Machine Learning and Knowledge Discovery in Databases: Part I
First-Order Multi-class Subgroup Discovery

Proceedings of the 2010 conference on STAIRS 2010: Proceedings of the Fifth Starting AI Researchers' Symposium
Learning multi-class theories in ILP

ILP'10 Proceedings of the 20th international conference on Inductive logic programming
Reducing number of classifiers in DAGSVM based on class similarity

ICIAP'11 Proceedings of the 16th international conference on Image analysis and processing: Part I
Constructing multiclass learners from binary learners: a simple black-box analysis of the generalization errors

ALT'05 Proceedings of the 16th international conference on Algorithmic Learning Theory

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper presents a method, called Adaptive Directed Acyclic Graph (ADAG), to extend Support Vector Machines (SVMs) for multiclass classification. The ADAG is based on the previous approach, the Decision Directed Acyclic Graph (DDAG), and is designed to remedy some weakness of the DDAG caused by its structure. We prove that the expected accuracy of the ADAG is higher than that of the DDAG, and also empirically evaluate our approach by comparing the ADAG with the DDAG on two data sets.