Research map / Vision-language and multimodal / Cross-modal learning
Context-Aware Generative Imputation for Robust Multimodal Learning in Missing Modality Scenarios
Jungwon Choi, Juho Lee
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes a context-aware generative imputation method to reconstruct missing modalities so multimodal models stay robust when data modalities are absent.
Problem
Multimodal models degrade severely when data modalities are missing at deployment due to cost, privacy, or sensor failure; shared-representation and expert-routing approaches struggle when modality-specific information is absent.
Approach
Generative imputation of the missing modality conditioned on available context, motivated by the observation that the target representation for an unobserved modality is not unique (one-to-many mapping), so the method models a distribution rather than a single deterministic reconstruction.
Closest papers by meaning
- Do More Modalities Always Help? A Geometric Perspective on Missing-Modality Robustness (NeurIPS 2026 · similarity 0.89)
- Knowledge Bridger: Towards Training-Free Missing Modality Completion (CVPR 2025 · similarity 0.89)
- Mitigating Pervasive Modality Absence Through Multimodal Generalization and Refinement (AAAI 2025 · similarity 0.88)
- Flex-MoE: Modeling Arbitrary Modality Combination via the Flexible Mixture-of-Experts (NeurIPS 2024 · similarity 0.88)
- RedCore: Relative Advantage Aware Cross-Modal Representation Learning for Missing Modalities with Imbalanced Missing Rates (AAAI 2024 · similarity 0.88)
- Unity by Diversity: Improved Representation Learning for Multimodal VAEs (NeurIPS 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
