Multimodal-Infomax
This repository contains the official implementation code of the paper Improving Multimodal Fusion with Hierarchical Mutual Information Maximization for Multimodal Sentiment Analysis, accepted at EMNLP 2021.
About
Multimodal-Infomax (MMIM) focuses on synthesizing fusion results from multi-modal inputs through a two-level mutual information maximization approach. It employs techniques such as the Barber-Agakov lower bound and contrastive predictive coding to enhance the training process. This software is designed for researchers and practitioners in the field of multimodal deep learning, particularly those interested in sentiment analysis.
Open Source Health
- Stars
- 203
- Forks
- 36
- License
- MIT
- Last commit
- 4 years ago
Resources & Links
Related Categories
Vendor
Deep Cognition and Language Research (DeCLaRe) Lab
Open-source projects on GitHub: Conv Emotion, Tango and MELD
Quick Links
Open Source
More by Deep Cognition and Language Research (DeCLaRe) Lab
Related Products
Twitter Sentiment Visualisation
🌍 Sentiment analysis over real-time social media data, rendering live charts to visualise trends
Top category match
Linusrants
Dataset of Linus Torvalds' rants classified by negativity using sentiment analysis
Top category match
PyABSA
Sentiment Analysis, Text Classification, Text Augmentation, Text Adversarial defense, etc.;
Top category match
Sentiment Discovery
Unsupervised Language Modeling at scale for robust sentiment classification
Top category match
LSTM-Sentiment-Analysis
Sentiment Analysis with LSTMs in Tensorflow
Top category match