Multimodal Information Interactions Bottleneck for Multimodal Self-Supervised Learning
cs.CV, cs.LG
Submitted: 2026-11-05
Updated: 2026-11-05
Code: https://github.com/wanglingzaishi/MIIB
Terminology
Sources
- Nonnegative Decomposition of Multivariate Information
- Dependency, Compression, and Synergy: A Unified Information-Theoretic View of Multimodal Learning
- Multimodal Deep Learning
- On the Benefits of Early Fusion in Multimodal Representation Learning
- A Cookbook of Self-Supervised Learning
- The information bottleneck method
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs