Loading Now

Summary of Envisioning Outlier Exposure by Large Language Models For Out-of-distribution Detection, By Chentao Cao et al.


Envisioning Outlier Exposure by Large Language Models for Out-of-Distribution Detection

by Chentao Cao, Zhun Zhong, Zhanke Zhou, Yang Liu, Tongliang Liu, Bo Han

First submitted to arxiv on: 2 Jun 2024

Categories

  • Main: Machine Learning (cs.LG)
  • Secondary: None

     Abstract of paper      PDF of paper


GrooveSquid.com Paper Summaries

GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!

Summary difficulty Written by Summary
High Paper authors High Difficulty Summary
Read the original abstract here
Medium GrooveSquid.com (original content) Medium Difficulty Summary
The proposed paper tackles the challenge of zero-shot out-of-distribution (OOD) sample detection using vision-language models like CLIP. Existing methods rely on text-based classifiers with closed-set labels, limiting the capability to recognize samples from large and open label spaces. The authors propose Envisioning Potential Outlier Exposure (EOE), leveraging large language models’ expert knowledge and reasoning capabilities without access to actual OOD data. EOE can be generalized to different tasks, including far, near, and fine-grained OOD detection. To achieve this, the authors design LLM prompts based on visual similarity to generate potential outlier class labels and a new score function based on potential outlier penalty to distinguish hard OOD samples effectively.
Low GrooveSquid.com (original content) Low Difficulty Summary
The paper proposes Envisioning Potential Outlier Exposure (EOE), which uses large language models’ expert knowledge and reasoning capabilities without access to actual out-of-distribution (OOD) data. This method can be used for different tasks, including far, near, and fine-grained OOD detection. The authors also design prompts based on visual similarity and a new score function to help distinguish hard OOD samples.

Keywords

» Artificial intelligence  » Zero shot