A maximum entropy framework for semisupervised and active learning with unknown and label-scarce classes

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

We investigate semisupervised learning (SL) and pool-based active learning (AL) of a classifier for domains with label-scarce (LS) and unknown categories, i.e., defined categories for which there are initially no labeled examples. This scenario manifests, e.g., when a category is rare, or expensive to label. There are several learning issues when there are unknown categories: 1) it is a priori unknown which subset of (possibly many) measured features are needed to discriminate unknown from common classes and 2) label scarcity suggests that overtraining is a concern. Our classifier exploits the inductive bias that an unknown class consists of the subset of the unlabeled pool's samples that are atypical (relative to the common classes) with respect to certain key (albeit a priori unknown) features and feature interactions. Accordingly, we treat negative log- p -values on raw features as nonnegatively weighted derived feature inputs to our class posterior, with zero weights identifying irrelevant features. Through a hierarchical class posterior, our model accommodates multiple common classes, multiple LS classes, and unknown classes. For learning, we propose a novel semisupervised objective customized for the LS/unknown category scenarios. While several works minimize class decision uncertainty on unlabeled samples, we instead preserve this uncertainty [maximum entropy (maxEnt)] to avoid overtraining. Our experiments on a variety of UCI Machine learning (ML) domains show: 1) the use of p -value features coupled with weight constraints leads to sparse solutions and gives significant improvement over the use of raw features and 2) for LS SL and AL, unlabeled samples are helpful, and should be used to preserve decision uncertainty (maxEnt), rather than to minimize it, especially during the early stages of AL. Our AL system, leveraging a novel sample-selection scheme, discovers unknown classes and discriminates LS classes from common ones, with sparing use of oracle labeling.

Original languageEnglish (US)
Article number7393573
Pages (from-to)917-933
Number of pages17
JournalIEEE Transactions on Neural Networks and Learning Systems
Volume28
Issue number4
DOIs
StatePublished - Apr 2017

Fingerprint

Labels
Entropy
Learning systems
Classifiers
Problem-Based Learning
Labeling
Uncertainty
Experiments

All Science Journal Classification (ASJC) codes

  • Software
  • Computer Science Applications
  • Computer Networks and Communications
  • Artificial Intelligence

Cite this

@article{df145699442f49dc8345eef2fbda38e1,
title = "A maximum entropy framework for semisupervised and active learning with unknown and label-scarce classes",
abstract = "We investigate semisupervised learning (SL) and pool-based active learning (AL) of a classifier for domains with label-scarce (LS) and unknown categories, i.e., defined categories for which there are initially no labeled examples. This scenario manifests, e.g., when a category is rare, or expensive to label. There are several learning issues when there are unknown categories: 1) it is a priori unknown which subset of (possibly many) measured features are needed to discriminate unknown from common classes and 2) label scarcity suggests that overtraining is a concern. Our classifier exploits the inductive bias that an unknown class consists of the subset of the unlabeled pool's samples that are atypical (relative to the common classes) with respect to certain key (albeit a priori unknown) features and feature interactions. Accordingly, we treat negative log- p -values on raw features as nonnegatively weighted derived feature inputs to our class posterior, with zero weights identifying irrelevant features. Through a hierarchical class posterior, our model accommodates multiple common classes, multiple LS classes, and unknown classes. For learning, we propose a novel semisupervised objective customized for the LS/unknown category scenarios. While several works minimize class decision uncertainty on unlabeled samples, we instead preserve this uncertainty [maximum entropy (maxEnt)] to avoid overtraining. Our experiments on a variety of UCI Machine learning (ML) domains show: 1) the use of p -value features coupled with weight constraints leads to sparse solutions and gives significant improvement over the use of raw features and 2) for LS SL and AL, unlabeled samples are helpful, and should be used to preserve decision uncertainty (maxEnt), rather than to minimize it, especially during the early stages of AL. Our AL system, leveraging a novel sample-selection scheme, discovers unknown classes and discriminates LS classes from common ones, with sparing use of oracle labeling.",
author = "Zhicong Qiu and Miller, {David J.} and George Kesidis",
year = "2017",
month = "4",
doi = "10.1109/TNNLS.2016.2514401",
language = "English (US)",
volume = "28",
pages = "917--933",
journal = "IEEE Transactions on Neural Networks and Learning Systems",
issn = "2162-237X",
publisher = "IEEE Computational Intelligence Society",
number = "4",

}

TY - JOUR

T1 - A maximum entropy framework for semisupervised and active learning with unknown and label-scarce classes

AU - Qiu, Zhicong

AU - Miller, David J.

AU - Kesidis, George

PY - 2017/4

Y1 - 2017/4

N2 - We investigate semisupervised learning (SL) and pool-based active learning (AL) of a classifier for domains with label-scarce (LS) and unknown categories, i.e., defined categories for which there are initially no labeled examples. This scenario manifests, e.g., when a category is rare, or expensive to label. There are several learning issues when there are unknown categories: 1) it is a priori unknown which subset of (possibly many) measured features are needed to discriminate unknown from common classes and 2) label scarcity suggests that overtraining is a concern. Our classifier exploits the inductive bias that an unknown class consists of the subset of the unlabeled pool's samples that are atypical (relative to the common classes) with respect to certain key (albeit a priori unknown) features and feature interactions. Accordingly, we treat negative log- p -values on raw features as nonnegatively weighted derived feature inputs to our class posterior, with zero weights identifying irrelevant features. Through a hierarchical class posterior, our model accommodates multiple common classes, multiple LS classes, and unknown classes. For learning, we propose a novel semisupervised objective customized for the LS/unknown category scenarios. While several works minimize class decision uncertainty on unlabeled samples, we instead preserve this uncertainty [maximum entropy (maxEnt)] to avoid overtraining. Our experiments on a variety of UCI Machine learning (ML) domains show: 1) the use of p -value features coupled with weight constraints leads to sparse solutions and gives significant improvement over the use of raw features and 2) for LS SL and AL, unlabeled samples are helpful, and should be used to preserve decision uncertainty (maxEnt), rather than to minimize it, especially during the early stages of AL. Our AL system, leveraging a novel sample-selection scheme, discovers unknown classes and discriminates LS classes from common ones, with sparing use of oracle labeling.

AB - We investigate semisupervised learning (SL) and pool-based active learning (AL) of a classifier for domains with label-scarce (LS) and unknown categories, i.e., defined categories for which there are initially no labeled examples. This scenario manifests, e.g., when a category is rare, or expensive to label. There are several learning issues when there are unknown categories: 1) it is a priori unknown which subset of (possibly many) measured features are needed to discriminate unknown from common classes and 2) label scarcity suggests that overtraining is a concern. Our classifier exploits the inductive bias that an unknown class consists of the subset of the unlabeled pool's samples that are atypical (relative to the common classes) with respect to certain key (albeit a priori unknown) features and feature interactions. Accordingly, we treat negative log- p -values on raw features as nonnegatively weighted derived feature inputs to our class posterior, with zero weights identifying irrelevant features. Through a hierarchical class posterior, our model accommodates multiple common classes, multiple LS classes, and unknown classes. For learning, we propose a novel semisupervised objective customized for the LS/unknown category scenarios. While several works minimize class decision uncertainty on unlabeled samples, we instead preserve this uncertainty [maximum entropy (maxEnt)] to avoid overtraining. Our experiments on a variety of UCI Machine learning (ML) domains show: 1) the use of p -value features coupled with weight constraints leads to sparse solutions and gives significant improvement over the use of raw features and 2) for LS SL and AL, unlabeled samples are helpful, and should be used to preserve decision uncertainty (maxEnt), rather than to minimize it, especially during the early stages of AL. Our AL system, leveraging a novel sample-selection scheme, discovers unknown classes and discriminates LS classes from common ones, with sparing use of oracle labeling.

UR - http://www.scopus.com/inward/record.url?scp=84961370132&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84961370132&partnerID=8YFLogxK

U2 - 10.1109/TNNLS.2016.2514401

DO - 10.1109/TNNLS.2016.2514401

M3 - Article

C2 - 26829808

AN - SCOPUS:84961370132

VL - 28

SP - 917

EP - 933

JO - IEEE Transactions on Neural Networks and Learning Systems

JF - IEEE Transactions on Neural Networks and Learning Systems

SN - 2162-237X

IS - 4

M1 - 7393573

ER -