@inproceedings{he-etal-2022-cpl,
title = "{CPL}: Counterfactual Prompt Learning for Vision and Language Models",
author = "He, Xuehai and
Yang, Diji and
Feng, Weixi and
Fu, Tsu-Jui and
Akula, Arjun and
Jampani, Varun and
Narayana, Pradyumna and
Basu, Sugato and
Wang, William Yang and
Wang, Xin",
editor = "Goldberg, Yoav and
Kozareva, Zornitsa and
Zhang, Yue",
booktitle = "Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2022",
address = "Abu Dhabi, United Arab Emirates",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2022.emnlp-main.224",
doi = "10.18653/v1/2022.emnlp-main.224",
pages = "3407--3418",
abstract = "Prompt tuning is a new few-shot transfer learning technique that only tunes the learnable prompt for pre-trained vision and language models such as CLIP. However, existing prompt tuning methods tend to learn spurious or entangled representations, which leads to poor generalization to unseen concepts. Towards non-spurious and efficient prompt learning from limited examples, this paper presents a novel Counterfactual Prompt Learning (CPL) method for vision and language models, which simultaneously employs counterfactual generation and contrastive learning in a joint optimization framework. Particularly, CPL constructs counterfactual by identifying minimal non-spurious feature change between semantically-similar positive and negative samples that causes concept change, and learns more generalizable prompt representation from both factual and counterfactual examples via contrastive learning. Extensive experiments demonstrate that CPL can obtain superior few-shot performance on different vision and language tasks than previous prompt tuning methods on CLIP. On image classification, we achieve 3.55{\%} average relative improvement on unseen classes across seven datasets; on image-text retrieval and visual question answering, we gain up to 4.09{\%} and 25.08{\%} relative improvements across three few-shot scenarios on unseen test sets respectively.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="he-etal-2022-cpl">
<titleInfo>
<title>CPL: Counterfactual Prompt Learning for Vision and Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Xuehai</namePart>
<namePart type="family">He</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Diji</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Weixi</namePart>
<namePart type="family">Feng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tsu-Jui</namePart>
<namePart type="family">Fu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Arjun</namePart>
<namePart type="family">Akula</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Varun</namePart>
<namePart type="family">Jampani</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pradyumna</namePart>
<namePart type="family">Narayana</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sugato</namePart>
<namePart type="family">Basu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">William</namePart>
<namePart type="given">Yang</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xin</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2022-12</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yoav</namePart>
<namePart type="family">Goldberg</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zornitsa</namePart>
<namePart type="family">Kozareva</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Abu Dhabi, United Arab Emirates</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Prompt tuning is a new few-shot transfer learning technique that only tunes the learnable prompt for pre-trained vision and language models such as CLIP. However, existing prompt tuning methods tend to learn spurious or entangled representations, which leads to poor generalization to unseen concepts. Towards non-spurious and efficient prompt learning from limited examples, this paper presents a novel Counterfactual Prompt Learning (CPL) method for vision and language models, which simultaneously employs counterfactual generation and contrastive learning in a joint optimization framework. Particularly, CPL constructs counterfactual by identifying minimal non-spurious feature change between semantically-similar positive and negative samples that causes concept change, and learns more generalizable prompt representation from both factual and counterfactual examples via contrastive learning. Extensive experiments demonstrate that CPL can obtain superior few-shot performance on different vision and language tasks than previous prompt tuning methods on CLIP. On image classification, we achieve 3.55% average relative improvement on unseen classes across seven datasets; on image-text retrieval and visual question answering, we gain up to 4.09% and 25.08% relative improvements across three few-shot scenarios on unseen test sets respectively.</abstract>
<identifier type="citekey">he-etal-2022-cpl</identifier>
<identifier type="doi">10.18653/v1/2022.emnlp-main.224</identifier>
<location>
<url>https://aclanthology.org/2022.emnlp-main.224</url>
</location>
<part>
<date>2022-12</date>
<extent unit="page">
<start>3407</start>
<end>3418</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T CPL: Counterfactual Prompt Learning for Vision and Language Models
%A He, Xuehai
%A Yang, Diji
%A Feng, Weixi
%A Fu, Tsu-Jui
%A Akula, Arjun
%A Jampani, Varun
%A Narayana, Pradyumna
%A Basu, Sugato
%A Wang, William Yang
%A Wang, Xin
%Y Goldberg, Yoav
%Y Kozareva, Zornitsa
%Y Zhang, Yue
%S Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing
%D 2022
%8 December
%I Association for Computational Linguistics
%C Abu Dhabi, United Arab Emirates
%F he-etal-2022-cpl
%X Prompt tuning is a new few-shot transfer learning technique that only tunes the learnable prompt for pre-trained vision and language models such as CLIP. However, existing prompt tuning methods tend to learn spurious or entangled representations, which leads to poor generalization to unseen concepts. Towards non-spurious and efficient prompt learning from limited examples, this paper presents a novel Counterfactual Prompt Learning (CPL) method for vision and language models, which simultaneously employs counterfactual generation and contrastive learning in a joint optimization framework. Particularly, CPL constructs counterfactual by identifying minimal non-spurious feature change between semantically-similar positive and negative samples that causes concept change, and learns more generalizable prompt representation from both factual and counterfactual examples via contrastive learning. Extensive experiments demonstrate that CPL can obtain superior few-shot performance on different vision and language tasks than previous prompt tuning methods on CLIP. On image classification, we achieve 3.55% average relative improvement on unseen classes across seven datasets; on image-text retrieval and visual question answering, we gain up to 4.09% and 25.08% relative improvements across three few-shot scenarios on unseen test sets respectively.
%R 10.18653/v1/2022.emnlp-main.224
%U https://aclanthology.org/2022.emnlp-main.224
%U https://doi.org/10.18653/v1/2022.emnlp-main.224
%P 3407-3418
Markdown (Informal)
[CPL: Counterfactual Prompt Learning for Vision and Language Models](https://aclanthology.org/2022.emnlp-main.224) (He et al., EMNLP 2022)
ACL
- Xuehai He, Diji Yang, Weixi Feng, Tsu-Jui Fu, Arjun Akula, Varun Jampani, Pradyumna Narayana, Sugato Basu, William Yang Wang, and Xin Wang. 2022. CPL: Counterfactual Prompt Learning for Vision and Language Models. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 3407–3418, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.