Adapting Large Multimodal Models to Distribution Shifts: The Role of In-Context Learning

TL;DR

This paper explores in-context learning for adapting large multimodal models to distribution shifts, proposing a novel method that improves example selection and significantly enhances model performance in domain-specific tasks.

Contribution

It introduces InvariantSelectPR, a new approach leveraging Class-conditioned Contrastive Invariance to improve demonstration selection for better adaptation of LMMs.

Findings

InvariantSelectPR improves accuracy by 34.2% on Camelyon17.

It achieves a 16.9% accuracy increase on HAM10000.

The method enhances the robustness of LMMs under distribution shifts.

Abstract

Recent studies indicate that large multimodal models (LMMs) potentially act as general-purpose assistants and are highly robust against different distributions. Despite this, domain-specific adaptation is still necessary particularly in specialized areas like healthcare. Due to the impracticality of fine-tuning LMMs given their vast parameter space, this work investigates in-context learning (ICL) as an effective alternative for enhancing LMMs' adaptability. Our study addresses this by evaluating an unsupervised ICL method which selects in-context examples through a nearest example search based on feature similarity. We uncover that its effectiveness is limited by the deficiencies of pre-trained vision encoders under distribution shift scenarios. To address these challenges, we propose InvariantSelectPR, a novel method leveraging Class-conditioned Contrastive Invariance (CCI) for more robust demonstration selection. Specifically, CCI enhances pre-trained vision encoders by improving their discriminative capabilities across different classes and ensuring invariance to domain-specific variations. This enhancement allows the encoders to effectively identify and retrieve the most informative examples, which are then used to guide LMMs in adapting to new query samples under varying distributions. Our experiments show that InvariantSelectPR substantially improves the adaptability of LMMs, achieving significant performance gains on benchmark datasets, with a 34.2%$\uparrow$ accuracy increase in 7-shot on Camelyon17 and 16.9%$\uparrow$ increase in 7-shot on HAM10000 compared to the baseline zero-shot performance.