Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
AngelWings1997 24500cf674 | 1 year ago | |
---|---|---|
.. | ||
Dataset | 1 year ago | |
presentation | 1 year ago | |
DRML_demo_CUB.py | 1 year ago | |
MTSVRC_dataset_process.py | 1 year ago | |
MTSVRC_run.py | 1 year ago | |
README.md | 1 year ago | |
evaluation.py | 1 year ago | |
evaluation_mAP.py | 1 year ago | |
mAP.py | 1 year ago | |
measure.py | 1 year ago | |
my_evalution.py | 1 year ago | |
tmp.py | 1 year ago |
This repository contains code for our AAAI Conference on Artificial Intelligence 2020 paper: Dual Relation Semi-supervised Multi-label Learning (DRML).
Multi-label learning (MLL) solves the problem that one single sample corresponds to multiple labels. It is a challenging task due to the long-tail label distribution and the sophisticated label relations. Semi-supervised MLL methods utilize a small-scale labeled samples and large-scale unlabeled samples to enhance the performance. However, these approaches mainly focus on exploring the data distribution in feature space while ignoring mining the label relation inside of each instance.
We proposed a Dual Relation Semi-supervised Multi-label Learning (DRML) approach which jointly explores the feature distribution and the label relation simultaneously. A dual-classifier domain adaptation strategy is proposed to align features while generating pseudo labels to improve learning performance. A relation network is proposed to explore the relation knowledge. As a result, DRML effectively explores the feature-label and label-label relations in both labeled and unlabeled samples. It is an end-to-end model without any extra knowledge. Extensive experiments illustrate the effectiveness and efficiency of our method.
DRML includes a novel domain adaptation co-training strategy and a label relation mining module in semi-supervised fashion. It explores both the instance similarity in feature space and the label-label relation in label space simultaneously. Specifically, deploy a two-classifier domain adaptation strategy to align the feature distribution in a latent space. Moreover, it further provides the pseudo label of unlabeled samples to enhance the training performance. Furthermore, a relation network is proposed to utilize the predictions from the two classifiers to learn the label relations. All modules are simultaneously optimized in an end-to-end manner to achieve the highest performance. The major contributions of our work are briefly listed as follows:
We combine all components of our model in a single python file. We run our code in Anaconda environment. We use Tensorflow and please see the code for the detailed package information. After config the environment, input the command below for the demo:
python DRML_demo_CUB.py
The demo loads the CUB dataset for training and evaluation. If you cannot download the datasets, please try the another Dataset Download link here for all 6 datasets of our experiments.
Welcome to send us Emails if you have any questions about the code and our work :-)
Please cite our paper if you like or use our work for your research, thank you very much!
@inproceedings{DRML_AAAI20,
title={Dual Relation Semi-supervised Multi-label Learning},
author={Wang, Lichen and Liu, Yunyu and Qin, Can and Sun, Gan and Fu, Yun},
booktitle={Proceedings of AAAI Conference on Artificial Intelligence},
year={2020}
}
No Description
Unity3D Asset MATLAB other
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》