This is the source code of paper "Robust-MSA: Understanding the Impact of Modality Noise on Multimodal Sentiment Analysis", which is accepted by AAAI 2023 Demonstration Program. The Demonstration Video can be viewed on Youtube
This demonstration system is developed based on the B/S architecture.
The front-end source code is located in the vue3 folder, while the published web pages can be found in the vue3/dist folder. They can be deployed using server softwares such as Nginx. There are two urls need to be adjusted in dist/config.js according to the IP address of the server.
The back-end is coded in Python, the package dependencies are listed in requirements.txt. Note that the MMSA-FET package needs additional post-installation setups to function properly as documented here. Server ports can be adjusted in config.py.
The noise wav files and the trained models can be downloaded from Google Drive or Baidu Drive.
If you find this work helpful, please cite us:
@article{mao2022robust,
title={Robust-MSA: Understanding the Impact of Modality Noise on Multimodal Sentiment Analysis},
author={Mao, Huisheng and Zhang, Baozheng and Xu, Hua and Yuan, Ziqi and Liu, Yihe},
journal={arXiv preprint arXiv:2211.13484},
year={2022}
}
@article{yuan2023noise,
title={Noise Imitation Based Adversarial Training for Robust Multimodal Sentiment Analysis},
author={Yuan, Ziqi and Liu, Yihe and Xu, Hua and Gao, Kai},
journal={IEEE Transactions on Multimedia},
year={2023},
publisher={IEEE}
}