This is my first journal work for my PhD studying, which is about Multi-label classification tasks. This work is based on the software pakeage ML-FE1.12 from [1].
This project is based on linear discrimiant analysis and a probabilistic saliency estimation model to build a novel framework for multi-label classification tasks. The framework can be utilized by various prior information of input data. Due to the calculation of weight factors is based on each class, the framework can allievate the imbalance problems existing in multi-label datasets. The paper was published in IEEE Transactions on Cybernetics as: Paper link
Matlab
Run this code under as follow: ./Validation/test_dataset.m
Choose the transform type according to comments and your requirements
The code of our work is based on the competed work 'A weighted linear discriminant analysis framework for multi-label feature extraction'.
[1] The competed and backbone work of our publication is provided as Paper link and Code link
This project is licensed under the MIT License - see the LICENSE.md file for details
Remember to cite our paper and work [1], if you would refer this work.
L. Xu, J. Raitoharju, A. Iosifidis and M. Gabbouj, "Saliency-Based Multilabel Linear Discriminant Analysis," in IEEE Transactions on Cybernetics, doi: 10.1109/TCYB.2021.3069338.