Papers/2609.21018
🧪 Test?View on arXiv

MAGIC: Marginal-Guided Compression with Optimal Transport for Efficient Visual Document Retrieval

Xander Y. Geek, Author 2, Author 3, Author 4, Author 5

compressionvisual document retrievaloptimal transportpost-hoc methods
2609.21018
Builder Relevance
80%
1h ago

Abstract

MAGIC proposes a training-free post-hoc compressor for efficient visual document retrieval that optimizes patch usage through a two-marginal entropic optimal-transport formulation.

Reality Card

Core Claim

MAGIC consistently outperforms strong post-hoc compressors across various benchmarks, particularly in aggressive-compression scenarios.

Method / Result

MAGIC achieves significant performance improvements in retrieval efficiency, especially under aggressive compression conditions.

Limitations

The method's effectiveness may vary based on specific retrieval backbones and benchmarks used.

Paper to code

Verified implementation resources so builders can test the paper’s claims instead of stopping at the abstract.

No verified implementation link has been attached yet. AIBuzzHub will keep this panel separate from unverified search results.
← Back to all papers