Metraipolis › archive › dc_741ff3362497

MiCo: Mutual Information Coverage Optimization through Semantic Erasure Modeling for Efficient MLLM Inference — arXiv:2609.34330

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.34330
License (as archived)
CC0-1.0
Archived
2026-10-07T23:52:02.739Z
SHA-256 of the archived text
3012754b301928e78563d8f2e27118233079d7e06ca7a1bdb28b06aab3a8fcb2
Stable URL of this page
https://metraipolis.com/documento/dc_741ff3362497

How to cite

MiCo: Mutual Information Coverage Optimization through Semantic Erasure Modeling for Efficient MLLM Inference — arXiv:2609.34330 — arXiv (en), https://arxiv.org/abs/2609.34330 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_741ff3362497 (Metraipolis archive dc_741ff3362497, sha256 3012754b301928e78563d8f2e27118233079d7e06ca7a1bdb28b06aab3a8fcb2)
BibTeX
@misc{dc741ff3362497,
  title = {MiCo: Mutual Information Coverage Optimization through Semantic Erasure Modeling for Efficient MLLM Inference — arXiv:2609.34330},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.34330},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_741ff3362497; sha256 3012754b301928e78563d8f2e27118233079d7e06ca7a1bdb28b06aab3a8fcb2},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.34330. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Multimodal large language models (MLLMs) have demonstrated impressive performance in multimodal understanding, but processing large numbers of visual tokens results in high computational costs. While many methods have been proposed to reduce the number of visual tokens, most of them rely on heuristics and are prone to discarding substantial visual information during pruning, leading to degradation in model performance. In this work, by using a semantic erasure model, we derive a general mutual information coverage objective from task log-loss and propose MiCo, a training-free two-stage pruning method. MiCo first uses visual signals to select a representative candidate pool before visual tokens enter the language model, then performs task-aware subset selection within it. At each stage, suitable observable proxies instantiate the derived objective as a monotone submodular coverage function, which MiCo greedily optimizes under the token budget. MiCo is evaluated on diverse MLLMs ranging from 7B to 13B parameters across a broad range of image and video benchmarks spanning general visual reasoning, fine-grained OCR and grounding, hallucination detection, and long-video understanding. MiCo consistently achieves the best performance across nearly all evaluated models under all pruning ratios. On LLaVA-NEXT-13B, MiCo uses only 5.6% visual tokens, retains 97.5% of baseline performance, and achieves a 3.8-fold inference speedup. Our experiments demonstrate the effectiveness of MiCo and our mutual information coverage objective for visual token pruning.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-29 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.34330 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-07T23:29:27.353Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).