Multi-objective evolutionary methods for hierarchical and multi-label classification

• Grant number: 16/50457-5
• Support Opportunities: Regular Research Grants
• Start date: June 01, 2017
• End date: May 31, 2019
• Field of knowledge: Physical Sciences and Mathematics - Computer Science
• Agreement: University of Surrey
• Mobility Program: SPRINT - Projetos de pesquisa - Mobilidade
• Principal Investigator: Ricardo Cerri
• Grantee: Ricardo Cerri
• Principal researcher abroad: Yaochu Jin
• Institution abroad: University of Surrey , England
• Host Institution: Centro de Ciências Exatas e de Tecnologia (CCET). Universidade Federal de São Carlos (UFSCAR). São Carlos , SP, Brazil
• City of the host institution: São Carlos
• Associated research grant: 15/14300-1 - Hierarchical classification of transposable elements using machine learning, AP.R

Abstract

In conventional classification, an instance is classified in just one among two or more classes. These problems are called single-label classification problems. However, there are more complex problems in which an instance can be classified in two or more classes simultaneously. These are known in the Machine Learning literature as multi-label classification problems. When the classes involved are organized in a hierarchy, the task is even more challenging, and is known as hierarchical classification. Many practical applications are related to hierarchical and multi-label classification, like the classification of images, documents and music, and protein function prediction. These are very challenging tasks due the difficulty in considering the relationships between the many classes of the problem during training. Together with this, the unbalance of the datasets harm the performances of the classifiers proposed in the literature. Given the high space of possible classes, we propose the development of Multi-objective Evolutionary Methods for the generation of multi-label and hierarchical classification rules. The rules generated should satisfy the requisites of interpretability and good performance. The proposed methods will be compared with state-of-the-art methods in the literature, using hierarchical and multi-label datasets from the bioinformatics domain. The method will be evaluated with measures specifically proposed for hierarchical and multi-label problems. (AU)

Scientific publications

(4)

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

PEREIRA, GEAN TRINDADE; GABRIEL, PAULO H. R.; CERRI, RICARDO; OLIVEIRA, PM; NOVAIS, P; REIS, LP. Hierarchical Classification of Transposable Elements with a Weighted Genetic Algorithm. PROGRESS IN ARTIFICIAL INTELLIGENCE, EPIA 2019, PT I, v. 11804, p. 13-pg., 2019-01-01. (16/50457-5, 15/14300-1)

MIRANDA, THIAGO ZAFALON; SARDINHA, DIORGE BROGNARA; CERRI, RICARDO. Preventing the generation of inconsistent sets of crisp classification rules. EXPERT SYSTEMS WITH APPLICATIONS, v. 165, MAR 1 2021. (16/50457-5)

PEREIRA, GEAN TRINDADE; GABRIEL, PAULO. H. R.; CERRI, RICARDO; LOPEZIBANEZ, M. A Lexicographic Genetic Algorithm for Hierarchical Classification Rule Induction. PROCEEDINGS OF THE 2019 GENETIC AND EVOLUTIONARY COMPUTATION CONFERENCE (GECCO'19), v. N/A, p. 9-pg., 2019-01-01. (15/14300-1, 16/50457-5)

CERRI, RICARDO; BASGALUPP, MARCIO P.; BARROS, RODRIGO C.; DE CARVALHO, ANDRE C. P. L. F.. Inducing Hierarchical Multi-label Classification rules with Genetic Algorithms. APPLIED SOFT COMPUTING, v. 77, p. 584-604, APR 2019. (16/50457-5, 15/14300-1)