Dependency, Compression, and Synergy: A Unified Information-Theoretic View of Multimodal Learning
cs.IT, cs.AI, cs.LG, math.IT
Submitted: 2026-09-13
Updated: 2026-09-13
Terminology
Sources
- Nonnegative Decomposition of Multivariate Information
- Representation Learning with Contrastive Predictive Coding
- OVA-IB: One vs All Information Bottleneck for Multi-Modal Alignment
- TVDIM: Enhancing Image Self-Supervised Pretraining via Noisy Text Data
- Orthogonalized Multimodal Contrastive Learning with Asymmetric Masking for Structured Representations
- SynIB: Informational Bottleneck for Maximizing Synergy in Multimodal Learning
- Multimodal Datasets with Controllable Mutual Information
- Zero-shot-Learning Cross-Modality Data Translation Through Mutual Information Guided Stochastic Diffusion
- Information-Theoretic Criteria for Knowledge Distillation in Multimodal Learning
- Distributed Information Bottleneck Theory for Multi-Modal Task-Aware Semantic Communication
- Towards Minimal Causal Representations for Human Multimodal Language Understanding
- Robust Multi-modal Task-oriented Communications with Redundancy-aware Representations
- Joint-Centric Dual Contrastive Alignment with Structure-Preserving and Information-Balanced Regularization
- Self-MI: Efficient Multimodal Fusion via Self-Supervised Multi-Task Learning with Auxiliary Mutual Information Maximization
- Multimodal Representations Learning Based on Mutual Information Maximization and Minimization and Identity Embedding for Multimodal Sentiment Analysis
- Balanced Multimodal Learning via Mutual Information
- Semantic Item Graph Enhancement for Multimodal Recommendation
- Enhancing Multimodal Recommendations with Vision-Language Models and Information-Aware Fusion
- Contribution-Guided Asymmetric Learning for Robust Multimodal Fusion under Imbalance and Noise
- OTCR: Optimal Transmission, Compression and Representation for Multimodal Information Extraction
Related papers
- Language Modeling is Monotone Compression
- Clipped Affine Policy: Low-Complexity Near-Optimal Online Power Control for Energy Harvesting Communications over Fading Channels
- Discrepancy for Random Linear Codes
- A New Approach to Code Smoothing Bounds
- Contextual Memory-Enhanced Source Coding for Low-SNR Communications
- Symmetry-Enforced Quadratic Approximate-Degradability Bounds for Noisy Landau-Streater Channels