Skip to content

Latest commit

 

History

History
76 lines (40 loc) · 4.58 KB

README.md

File metadata and controls

76 lines (40 loc) · 4.58 KB
Raw

Java 8 License

AutoMLC

Automated Multi-Label Classification

Overview

This project presents the Automated Multi-Label Classification (AutoMLC) project.

This is a project that provides 6 AutoML methods in the context of multi-label classification:

  1. GA-Auto-MLC -- Genetic Algorithm for Automated Multi-Label Classification
  2. Auto-MEKA_GGP -- Automated Multi-Label Classification with Grammar-based Genetic Programming
  3. Auto-MEKA_spGGP -- Automated Multi-Label Classification with Speciation-Grammar-based Genetic Programming
  4. AutoMEKA_BO -- Automated Multi-Label Classification with Bayesian Optimization
  5. Auto_MEKA_RS -- Automated Multi-Label Classification with Random Search
  6. Auto-MEKA_LS -- Automated Multi-Label Classification with Local Search

All methods try to enhance the performance of multi-label classification algorithms on the MEKA tool.

Publications

If you use this project in your work, be aware of the details in the following papers:

  1. GA-Auto-MLC:
  • A. G. C. de Sá, G. L. Pappa, and A. A. Freitas. Towards a method for automatically selecting and configuring multi-label classification algorithms. In Proceedings of the Genetic and Evolutionary Computation Conference Companion , pp. 1125–1132, 2017. [ PDF ] [ ACM ]
  1. Auto-MEKA_GGP:
  • A. G. C. de Sá, G. L. Pappa, and A. A. Freitas. Automated Selection and Configuration of Multi-Label Classification Algorithms with Grammar-based Genetic Programming. In Proceedings of the International Conference on Parallel Problem Solving from Nature (PPSN), 2018. [ PDF ] [ Springer ]
  1. Auto-MEKA_BO, Auto-MEKA_spGGP, Auto-MEKA_RS and Auto-MEKA_LS:
  • A. G. C. de Sá, C. G. Pimenta, A. A. Freitas, and G. L. Pappa. A robust experimental evaluation of automated multi-label classification methods. In Proceedings of the Genetic and Evolutionary Computation Conference, pp. 175–183, 2020. [ PDF ] [ ACM ]
  1. The description of the multi-label classification (MLC) search is available in the following report:
  • A. G. C. de Sá, C. G. Pimenta, A. A. Freitas, and G. L. Pappa. Multi-label classification search space in the MEKA software, 2020. [ arXiv ]
  1. Note that Auto-MEKA_GGP uses modified versions of EpochX, MEKA and WEKA to perform automated multi-label classification. The versions of these frameworks can be found at: EpochX, MEKA, and WEKA

How to Use the AutoML Methods

All methods depend on Java 8. We recommend you to use the jdk 1.8.0 281 as all methods were compiled on this jdk version. However, any jdk for Java 8 should be fine. You can find jdk for Java 8 to download at: https://www.oracle.com/java/technologies/javase/javase-jdk8-downloads.html

You are going to use the jdk directory to run the AutoML methods. You can find the details on how to run the AutoML methods for multi-label classification problems in the following links:

GA-Auto-MLC

Auto-MEKA_GGP

Auto-MEKA_spGGP

Auto-MEKA_BO

Auto-MEKA_RS

Auto-MEKA_LS

Datasets

Several datatasets are available at: datasets.

License

See LICENSE file.

Support

You should direct your question, doubt or comment to an issue on this repository. Although, you can also contact via email, an issue on github would also help others when they use this project. That's the main issue an issue is preferable.