Active learning - Approaches and issues

Tirthankar RayChaudhuri, Leonard G C Hamey

Research output: Contribution to journalArticleResearchpeer-review

Abstract

This paper surveys published work in active learning research with the purpose of providing a unified understanding of the area. A passive learning system relies entirely on pre-gathered information, whereas an active learning algorithm has the capability of interacting with its environment in order to collect information and/or to select learning policy. Active learning systems produce improved generalisation, reduce data costs and are most useful where data is expensive and computation is cheap. There are three major recognised approaches to the implementation of active learning - goal-driven learning, reinforcement learning and querying. While the first is largely a meta-level symbolic approach, the second is more a class of problems employing a policy-based approach to learning in non-deterministic dynamic environments; the third is based upon gathering the most useful examples by asking 'intelligent' questions. Research in the area is still mostly at a theoretical level.

LanguageEnglish
Pages205-243
Number of pages39
JournalJournal of Intelligent Systems
Volume7
Issue number3-4
Publication statusPublished - 1997

Fingerprint

Learning systems
Reinforcement learning
Learning algorithms
Problem-Based Learning
Costs

Keywords

  • Active Learning
  • Exploration and Exploitation
  • Goal-Driven Learning
  • Markov Decision Processes
  • Optimum Experiment Design
  • Query Filtering
  • Reinforcement Learning

Cite this

RayChaudhuri, Tirthankar ; Hamey, Leonard G C. / Active learning - Approaches and issues. In: Journal of Intelligent Systems. 1997 ; Vol. 7, No. 3-4. pp. 205-243.
@article{b538b091faa7437d8d760d5d9ea18a62,
title = "Active learning - Approaches and issues",
abstract = "This paper surveys published work in active learning research with the purpose of providing a unified understanding of the area. A passive learning system relies entirely on pre-gathered information, whereas an active learning algorithm has the capability of interacting with its environment in order to collect information and/or to select learning policy. Active learning systems produce improved generalisation, reduce data costs and are most useful where data is expensive and computation is cheap. There are three major recognised approaches to the implementation of active learning - goal-driven learning, reinforcement learning and querying. While the first is largely a meta-level symbolic approach, the second is more a class of problems employing a policy-based approach to learning in non-deterministic dynamic environments; the third is based upon gathering the most useful examples by asking 'intelligent' questions. Research in the area is still mostly at a theoretical level.",
keywords = "Active Learning, Exploration and Exploitation, Goal-Driven Learning, Markov Decision Processes, Optimum Experiment Design, Query Filtering, Reinforcement Learning",
author = "Tirthankar RayChaudhuri and Hamey, {Leonard G C}",
year = "1997",
language = "English",
volume = "7",
pages = "205--243",
journal = "Journal of Intelligent Systems",
issn = "0334-1860",
publisher = "De Gruyter",
number = "3-4",

}

RayChaudhuri, T & Hamey, LGC 1997, 'Active learning - Approaches and issues', Journal of Intelligent Systems, vol. 7, no. 3-4, pp. 205-243.

Active learning - Approaches and issues. / RayChaudhuri, Tirthankar; Hamey, Leonard G C.

In: Journal of Intelligent Systems, Vol. 7, No. 3-4, 1997, p. 205-243.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Active learning - Approaches and issues

AU - RayChaudhuri, Tirthankar

AU - Hamey, Leonard G C

PY - 1997

Y1 - 1997

N2 - This paper surveys published work in active learning research with the purpose of providing a unified understanding of the area. A passive learning system relies entirely on pre-gathered information, whereas an active learning algorithm has the capability of interacting with its environment in order to collect information and/or to select learning policy. Active learning systems produce improved generalisation, reduce data costs and are most useful where data is expensive and computation is cheap. There are three major recognised approaches to the implementation of active learning - goal-driven learning, reinforcement learning and querying. While the first is largely a meta-level symbolic approach, the second is more a class of problems employing a policy-based approach to learning in non-deterministic dynamic environments; the third is based upon gathering the most useful examples by asking 'intelligent' questions. Research in the area is still mostly at a theoretical level.

AB - This paper surveys published work in active learning research with the purpose of providing a unified understanding of the area. A passive learning system relies entirely on pre-gathered information, whereas an active learning algorithm has the capability of interacting with its environment in order to collect information and/or to select learning policy. Active learning systems produce improved generalisation, reduce data costs and are most useful where data is expensive and computation is cheap. There are three major recognised approaches to the implementation of active learning - goal-driven learning, reinforcement learning and querying. While the first is largely a meta-level symbolic approach, the second is more a class of problems employing a policy-based approach to learning in non-deterministic dynamic environments; the third is based upon gathering the most useful examples by asking 'intelligent' questions. Research in the area is still mostly at a theoretical level.

KW - Active Learning

KW - Exploration and Exploitation

KW - Goal-Driven Learning

KW - Markov Decision Processes

KW - Optimum Experiment Design

KW - Query Filtering

KW - Reinforcement Learning

UR - http://www.scopus.com/inward/record.url?scp=0031364797&partnerID=8YFLogxK

M3 - Article

VL - 7

SP - 205

EP - 243

JO - Journal of Intelligent Systems

T2 - Journal of Intelligent Systems

JF - Journal of Intelligent Systems

SN - 0334-1860

IS - 3-4

ER -