In my database, I have a significant number of labels (more than 10), the objective is to train an algorithm that, in response to the input values, gives me the response of the values of each label.
I would like to know what algorithm I could use and what software do you recommend? I was thinking of matlab.
I have used libraries for mullti-label classification in the following languages:
- Java -> "mulan" tool (http://mulan.sourceforge.net/)
- R -> "mldr" (https://cran.r-project.org/web/packages/mldr/index.html) and "utiml" packages ( https://cran.r-project.org/web/packages/utiml/index.html )
- Python -> scikit-learn package ( https://scikit-learn.org/stable/modules/multiclass.html)
Complementing Eduardo's answer.
There is also scikit-multilearn in Python:
http://scikit.ml/
and MEKA in Java:
http://waikato.github.io/meka/
In addition, if you want a very simple library to run Mulan, I would recommend this one:
https://github.com/kdis-lab/ExecuteMulan
However, I do not know anything in Matlab.
_______________________
Regarding the algorithms, I would recommend Binary Relevance (BR), Classifier Chain (CC), Hierarchy Of Multi-Label ClassiERs (HOMER) and Random Forest of Predictive Clustering Trees (RF-PCT).
Thank you very much everyone for responding, as Chinedu Pascal Ezenkwu says I have been able to do something using "Matlab Neural networks toolbox", however, I must validate other algorithms, so I will take your comments into account, I hope you can help me later if something happens .
thank you.
- Badges
- Science topic
- Similar topics
- Computer Science
- Data Mining