This script is the main training file for the AEMLO model. It handles the model initialization, training process, parameter tuning, and the final model saving.
This file is responsible for generating instances for the trained AEMLO.
This file contains the core model definition for the AEMLO project. It implements the model architecture, forward propagation, and backpropagation for deep learning-related functions.
This file includes preprocessing functions, such as data cleaning, feature extraction, and transformation. These functions prepare the input data for the model.
This file is a multi-label experimental module that integrates AEMLO-sampled instances with the original training set. It then applies well-known multi-label classifiers such as BR, MLkNN, and RAkEL to evaluate the model's performance. The results are displayed using various metrics, including Macro-level evaluations, providing a comprehensive comparison of the classifiers' effectiveness.
This is a Jupyter Notebook file used for interactively running the AEMLO with different base Multi-label Classifier.
Please, install the following packages
numpy 1.21.6
pytorch 1.11.0
scikit-learn 1.0.2
scikit-multilearn 0.2.0
and python==3.7
About scikit-multilearn reference scikit-multilearn GitHub Repository.
This package has many errors, such as iterative_train_test_split, you can refer to its issues for resolution.
run Main.ipynb
Regarding Parameters
parameters = {
"dataname": dataname,
"feat_dim": X[train].shape[1],
"num_labels": y[train].shape[1],
// The core dimension parameters can be understood by referring to the paper
"latent_dim": params["latent_dim"],
"fx_h_dim": params["fx_h_dim"],
"fe_h_dim": params["fe_h_dim"],
"fd_h_dim": params["fd_h_dim"],
"X_train": X[train],
"y_train": y[train],
"epoch": params["epoch"],
"batch_size": params["batch_size"],
"learningrate": params["learningrate"]
}If you find this code useful in your research, please consider citing us:
@inproceedings{zhou2024aemlo,
title={AEMLO: AutoEncoder-Guided Multi-label Oversampling},
author={Zhou, Ao and Liu, Bin and Wang, Jin and Sun, Kaiwei and Liu, Kelin},
booktitle={Joint European Conference on Machine Learning and Knowledge Discovery in Databases},
pages={107--124},
year={2024}
}If you have any specific questions, please contact zacqupt@gmail.com