I'm working with data that can belong to more than one output label.
I know a possible solution is to create a new label defined by the combination of one or more labels, but I'm looking for a classifier that is trained from a multiple label vector. Is there a linear classifier that can perform this task?
Hi Daniel Prado de Campos
If you have data with many labels, then you are dealing with a multi-label classification task. For this type of learning, you can use any linear algorithm (e.g., SVM with linear kernel ) and then wrap it into a multi-label classifier such as "Binary Relevance" algorithm.
HTH.
Dr. Samer Sarsam
you can plot your data in single one frame figure for multiple data visualizations
I have also searched a lot of this problem but you will train and test your data w.r.t each output label separately. Please Read this article.
This tutorial (in Portuguese) might be helpful:
https://www.researchgate.net/publication/329375616_Introducao_a_Classificacao_Multirrotulo
make algorithm, of multiple labeled and visualized them in a single script
If a data can be classified as part of diferent clases, it is posible to use fuzzy approaches. By using fuzzy logic, it is posible to assign a different membership value in terms of the similarity to a certain class.
Daniel Prado de Campos , you can make predictions using many machine learning algorithm , and data sets labeling and multiple outputs you can used the label segmentation of image
- Badges
- Science topic