@inproceedings{mussmann-etal-2020-importance,
title = "{O}n the {I}mportance of {A}daptive {D}ata {C}ollection for {E}xtremely {I}mbalanced {P}airwise {T}asks",
author = "Mussmann, Stephen and
Jia, Robin and
Liang, Percy",
editor = "Cohn, Trevor and
He, Yulan and
Liu, Yang",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2020",
month = nov,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2020.findings-emnlp.305",
doi = "10.18653/v1/2020.findings-emnlp.305",
pages = "3400--3413",
abstract = "Many pairwise classification tasks, such as paraphrase detection and open-domain question answering, naturally have extreme label imbalance (e.g., 99.99{\%} of examples are negatives). In contrast, many recent datasets heuristically choose examples to ensure label balance. We show that these heuristics lead to trained models that generalize poorly: State-of-the art models trained on QQP and WikiQA each have only 2.4{\%} average precision when evaluated on realistically imbalanced test data. We instead collect training data with active learning, using a BERT-based embedding model to efficiently retrieve uncertain points from a very large pool of unlabeled utterance pairs. By creating balanced training data with more informative negative examples, active learning greatly improves average precision to 32.5{\%} on QQP and 20.1{\%} on WikiQA.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="mussmann-etal-2020-importance">
<titleInfo>
<title>On the Importance of Adaptive Data Collection for Extremely Imbalanced Pairwise Tasks</title>
</titleInfo>
<name type="personal">
<namePart type="given">Stephen</namePart>
<namePart type="family">Mussmann</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Robin</namePart>
<namePart type="family">Jia</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Percy</namePart>
<namePart type="family">Liang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2020-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2020</title>
</titleInfo>
<name type="personal">
<namePart type="given">Trevor</namePart>
<namePart type="family">Cohn</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yulan</namePart>
<namePart type="family">He</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yang</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Online</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Many pairwise classification tasks, such as paraphrase detection and open-domain question answering, naturally have extreme label imbalance (e.g., 99.99% of examples are negatives). In contrast, many recent datasets heuristically choose examples to ensure label balance. We show that these heuristics lead to trained models that generalize poorly: State-of-the art models trained on QQP and WikiQA each have only 2.4% average precision when evaluated on realistically imbalanced test data. We instead collect training data with active learning, using a BERT-based embedding model to efficiently retrieve uncertain points from a very large pool of unlabeled utterance pairs. By creating balanced training data with more informative negative examples, active learning greatly improves average precision to 32.5% on QQP and 20.1% on WikiQA.</abstract>
<identifier type="citekey">mussmann-etal-2020-importance</identifier>
<identifier type="doi">10.18653/v1/2020.findings-emnlp.305</identifier>
<location>
<url>https://aclanthology.org/2020.findings-emnlp.305</url>
</location>
<part>
<date>2020-11</date>
<extent unit="page">
<start>3400</start>
<end>3413</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T On the Importance of Adaptive Data Collection for Extremely Imbalanced Pairwise Tasks
%A Mussmann, Stephen
%A Jia, Robin
%A Liang, Percy
%Y Cohn, Trevor
%Y He, Yulan
%Y Liu, Yang
%S Findings of the Association for Computational Linguistics: EMNLP 2020
%D 2020
%8 November
%I Association for Computational Linguistics
%C Online
%F mussmann-etal-2020-importance
%X Many pairwise classification tasks, such as paraphrase detection and open-domain question answering, naturally have extreme label imbalance (e.g., 99.99% of examples are negatives). In contrast, many recent datasets heuristically choose examples to ensure label balance. We show that these heuristics lead to trained models that generalize poorly: State-of-the art models trained on QQP and WikiQA each have only 2.4% average precision when evaluated on realistically imbalanced test data. We instead collect training data with active learning, using a BERT-based embedding model to efficiently retrieve uncertain points from a very large pool of unlabeled utterance pairs. By creating balanced training data with more informative negative examples, active learning greatly improves average precision to 32.5% on QQP and 20.1% on WikiQA.
%R 10.18653/v1/2020.findings-emnlp.305
%U https://aclanthology.org/2020.findings-emnlp.305
%U https://doi.org/10.18653/v1/2020.findings-emnlp.305
%P 3400-3413
Markdown (Informal)
[On the Importance of Adaptive Data Collection for Extremely Imbalanced Pairwise Tasks](https://aclanthology.org/2020.findings-emnlp.305) (Mussmann et al., Findings 2020)
ACL