Mutual Information Maximization

Mutual Information Maximization (MIM) aims to learn representations that capture the strongest statistical dependencies between different data modalities or aspects, thereby improving model performance in various machine learning tasks. Current research focuses on applying MIM to diverse problems, including self-supervised learning, generalized category discovery, and knowledge distillation, often employing contrastive learning frameworks and neural networks with specialized modules for information extraction and disentanglement. These advancements are significantly impacting fields like robotics, computer vision, and natural language processing by enabling more robust and efficient models, particularly in scenarios with limited labeled data or complex data structures.

Papers

May 25, 2022

Mutual Information Divergence: A Unified Metric for Multimodal Generative Models
Jin-Hwa Kim, Yunji Kim, Jiyoung Lee, Kang Min Yoo, Sang-Woo Lee
Text to Image Generation Mutual Information Multimodal Representation Learning Mutual Information Maximization Multimodal Generative Model

May 24, 2022

May 10, 2022

Deep Graph Clustering via Mutual Information Maximization and Mixture Model
Maedeh Ahmadi, Mehran Safayani, Abdolreza Mirzaei
Contrastive Learning Community Detection Mixture Model Graph Clustering Mutual Information Maximization Deep Graph Clustering Self Supervised Graph

April 20, 2022

Unsupervised Domain Adaptation for Cardiac Segmentation: Towards Structure Mutual Information Maximization
Changjie Lu, Shen Zheng, Gaurav Gupta
Domain Adaptation Unsupervised Domain Adaptation Domain Invariant Feature Cardiac Segmentation Mutual Information Maximization

March 29, 2022

Temporal Feature Alignment and Mutual Information Maximization for Video-Based Human Pose Estimation
Zhenguang Liu, Runyang Feng, Haoming Chen, Shuang Wu, Yixing Gao, Yunjun Gao, Xiang Wang
Pose Estimation Human Pose Estimation Temporal Alignment Mutual Information Maximization

March 23, 2022

Node Representation Learning in Graph via Node-to-Neighbourhood Mutual Information Maximization
Wei Dong, Junsheng Wu, Yi Luo, Zongyuan Ge, Peng Wang
Representation Learning Graph Drawing Negative Sampling Mutual Information Maximization Node Representation Learning

January 10, 2022

Multimodal Representations Learning Based on Mutual Information Maximization and Minimization and Identity Embedding for Multimodal Sentiment Analysis
Jiahao Zheng, Sen Zhang, Xiaoping Wang, Zhigang Zeng
Multimodal Representation Multimodal Sentiment Analysis Multimodal Representation Learning Multimodal Interaction Mutual Information Maximization Multimodal Joint

January 9, 2022

Preserving Domain Private Representation via Mutual Information Maximization
Jiahong Chen, Jing Wang, Weipeng Lin, Kuangen Zhang, Clarence W. de Silva
Unsupervised Domain Adaptation Domain Invariant Representation Mutual Information Maximization Cross Domain Generalization Privacy Preserving Representation Domain Divergence

Mutual Information Maximization

Papers

Mutual Information Divergence: A Unified Metric for Multimodal Generative Models

First Contact: Unsupervised Human-Machine Co-Adaptation via Mutual Information Maximization

SepIt: Approaching a Single Channel Speech Separation Bound

Deep Graph Clustering via Mutual Information Maximization and Mixture Model

Unsupervised Domain Adaptation for Cardiac Segmentation: Towards Structure Mutual Information Maximization

Temporal Feature Alignment and Mutual Information Maximization for Video-Based Human Pose Estimation

Node Representation Learning in Graph via Node-to-Neighbourhood Mutual Information Maximization

Multimodal Representations Learning Based on Mutual Information Maximization and Minimization and Identity Embedding for Multimodal Sentiment Analysis

Preserving Domain Private Representation via Mutual Information Maximization