Exploring Large Language Models for Multi-Modal Out-of-Distribution Detection
This work addresses the reliability of machine learning systems for applications requiring robust OOD detection, though it appears incremental as it builds on existing multi-modal approaches.
The paper tackles the problem of multi-modal out-of-distribution detection by addressing the issue of large language model hallucinations, proposing a selective generation method with consistency-based uncertainty calibration and visual object extraction, resulting in consistent outperformance over state-of-the-art methods.
Out-of-distribution (OOD) detection is essential for reliable and trustworthy machine learning. Recent multi-modal OOD detection leverages textual information from in-distribution (ID) class names for visual OOD detection, yet it currently neglects the rich contextual information of ID classes. Large language models (LLMs) encode a wealth of world knowledge and can be prompted to generate descriptive features for each class. Indiscriminately using such knowledge causes catastrophic damage to OOD detection due to LLMs' hallucinations, as is observed by our analysis. In this paper, we propose to apply world knowledge to enhance OOD detection performance through selective generation from LLMs. Specifically, we introduce a consistency-based uncertainty calibration method to estimate the confidence score of each generation. We further extract visual objects from each image to fully capitalize on the aforementioned world knowledge. Extensive experiments demonstrate that our method consistently outperforms the state-of-the-art.