Research map / Label-efficient and robust learning / Recognition and long-tail
Contrastive Learning with Data Misalignment: Feature Purity, Training Dynamics and Theoretical Generalization Guarantees
Read the paperCheck your idea against 108,967 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.
Provides the first theoretical analysis of contrastive learning on image-text pairs with data misalignment, covering feature purity, training dynamics, and generalization guarantees.
Problem
Contrastive learning from image-text pairs is widely used, but its theoretical foundations are underexplored when image-text pairs are misaligned.
Approach
Theoretical analysis showing that ground-truth modality-paired features are amplified during training while spurious features are suppressed under data misalignment, with generalization guarantees.
Key results
- The abstract states that ground-truth modality-paired features are amplified while spurious features are suppressed during contrastive training under misalignment; no numeric results are available from the abstract.
Closest papers by meaning
- On the Similarities of Embeddings in Contrastive Learning (ICML 2025 · similarity 0.90)
- A Statistical Theory of Contrastive Learning via Approximate Sufficient Statistics (NeurIPS 2025 · similarity 0.89)
- Your contrastive learning problem is secretly a distribution alignment problem (NeurIPS 2024 · similarity 0.89)
- Variational Supervised Contrastive Learning (NeurIPS 2025 · similarity 0.89)
- $\mathbb{X}$-Sample Contrastive Loss: Improving Contrastive Learning with Sample Similarity Graphs (ICLR 2025 · similarity 0.88)
- The "Law'' of the Unconscious Contrastive Learner: Probabilistic Alignment of Unpaired Modalities (ICLR 2025 · similarity 0.88)
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.