Learning transferable negative prompts for out-of-distribution detection
Existing prompt learning methods have shown certain capabilities in Out-of-Distribution (OOD) detection, but the lack of OOD images in the target dataset in their training can lead to mismatches between OOD images and In-Distribution (ID) categories, resulting in a high false positive rate. To addre...
Saved in:
Main Authors: | , , , , |
---|---|
Format: | text |
Language: | English |
Published: |
Institutional Knowledge at Singapore Management University
2024
|
Subjects: | |
Online Access: | https://ink.library.smu.edu.sg/sis_research/9759 https://ink.library.smu.edu.sg/context/sis_research/article/10759/viewcontent/2404.03248v1.pdf |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Institution: | Singapore Management University |
Language: | English |
id |
sg-smu-ink.sis_research-10759 |
---|---|
record_format |
dspace |
spelling |
sg-smu-ink.sis_research-107592024-12-16T02:54:19Z Learning transferable negative prompts for out-of-distribution detection LI, Tianqi PANG, Guansong BAI, Xiao MIAO, Wenjun ZHENG, Jin Existing prompt learning methods have shown certain capabilities in Out-of-Distribution (OOD) detection, but the lack of OOD images in the target dataset in their training can lead to mismatches between OOD images and In-Distribution (ID) categories, resulting in a high false positive rate. To address this issue, we introduce a novel OOD detection method, named ‘NegPrompt’, to learn a set of negative prompts, each representing a negative connotation of a given class label, for delineating the boundaries between ID and OOD images. It learns such negative prompts with ID data only, without any reliance on external out-lier data. Further, current methods assume the availability of samples of all ID classes, rendering them ineffective in open-vocabulary learning scenarios where the inference stage can contain novel ID classes not present during training. In contrast, our learned negative prompts are transferable to novel class labels. Experiments on various ImageNet benchmarks show that NegPrompt surpasses state-of-the-art prompt-learning-based OOD detection methods and maintains a consistent lead in hard OOD detection in closed- and open-vocabulary classification scenarios. 2024-06-01T07:00:00Z text application/pdf https://ink.library.smu.edu.sg/sis_research/9759 info:doi/10.1109/CVPR52733.2024.01665 https://ink.library.smu.edu.sg/context/sis_research/article/10759/viewcontent/2404.03248v1.pdf http://creativecommons.org/licenses/by-nc-nd/4.0/ Research Collection School Of Computing and Information Systems eng Institutional Knowledge at Singapore Management University Out-of-Distribution detection OOD detection Open-vocabulary learning Closed-vocabulary classification Open-vocabulary classification Artificial Intelligence and Robotics Computer Sciences |
institution |
Singapore Management University |
building |
SMU Libraries |
continent |
Asia |
country |
Singapore Singapore |
content_provider |
SMU Libraries |
collection |
InK@SMU |
language |
English |
topic |
Out-of-Distribution detection OOD detection Open-vocabulary learning Closed-vocabulary classification Open-vocabulary classification Artificial Intelligence and Robotics Computer Sciences |
spellingShingle |
Out-of-Distribution detection OOD detection Open-vocabulary learning Closed-vocabulary classification Open-vocabulary classification Artificial Intelligence and Robotics Computer Sciences LI, Tianqi PANG, Guansong BAI, Xiao MIAO, Wenjun ZHENG, Jin Learning transferable negative prompts for out-of-distribution detection |
description |
Existing prompt learning methods have shown certain capabilities in Out-of-Distribution (OOD) detection, but the lack of OOD images in the target dataset in their training can lead to mismatches between OOD images and In-Distribution (ID) categories, resulting in a high false positive rate. To address this issue, we introduce a novel OOD detection method, named ‘NegPrompt’, to learn a set of negative prompts, each representing a negative connotation of a given class label, for delineating the boundaries between ID and OOD images. It learns such negative prompts with ID data only, without any reliance on external out-lier data. Further, current methods assume the availability of samples of all ID classes, rendering them ineffective in open-vocabulary learning scenarios where the inference stage can contain novel ID classes not present during training. In contrast, our learned negative prompts are transferable to novel class labels. Experiments on various ImageNet benchmarks show that NegPrompt surpasses state-of-the-art prompt-learning-based OOD detection methods and maintains a consistent lead in hard OOD detection in closed- and open-vocabulary classification scenarios. |
format |
text |
author |
LI, Tianqi PANG, Guansong BAI, Xiao MIAO, Wenjun ZHENG, Jin |
author_facet |
LI, Tianqi PANG, Guansong BAI, Xiao MIAO, Wenjun ZHENG, Jin |
author_sort |
LI, Tianqi |
title |
Learning transferable negative prompts for out-of-distribution detection |
title_short |
Learning transferable negative prompts for out-of-distribution detection |
title_full |
Learning transferable negative prompts for out-of-distribution detection |
title_fullStr |
Learning transferable negative prompts for out-of-distribution detection |
title_full_unstemmed |
Learning transferable negative prompts for out-of-distribution detection |
title_sort |
learning transferable negative prompts for out-of-distribution detection |
publisher |
Institutional Knowledge at Singapore Management University |
publishDate |
2024 |
url |
https://ink.library.smu.edu.sg/sis_research/9759 https://ink.library.smu.edu.sg/context/sis_research/article/10759/viewcontent/2404.03248v1.pdf |
_version_ |
1819113130392813568 |