Active learning (machine learning): Difference between revisions

Revision as of 14:31, 31 August 2023

Revision as of 15:54, 31 August 2023

Line 3:

{{Machine learning bar}}

'''Active learning''' is a special case of [[machine learning]] in which a learning algorithm can interactively query a user (or some other information source) to label new data points with the desired outputs.<ref name="settles">{{cite web

| title = Active Learning Literature Survey

| url = http://pages.cs.wisc.edu/~bsettles/pub/settles.activelearning.pdf

| author = Settles, Burr

Line 48:

}}</ref> In statistics literature, it is sometimes also called [[optimal experimental design]].<ref name="olsson">{{cite web | url=http://eprints.sics.se/3600/ | title=A literature survey of active machine learning in the context of natural language processing |series=SICS Technical Report T2009:06 | author=Olsson, Fredrik| date=April 2009 }}</ref> The information source is also called ''teacher'' or ''oracle''.

Sometimes,There we come acrossare situations wherein wewhich have a lot ofunlabeled data withoutis any labels,abundant but it'smanual reallylabeling is expensive and time-consuming to label it all manually. In casessuch likea thisscenario, we can use special learning techniques.algorithms Thesecan techniquesactively involve the learning system askingquery the user or /teacher for labels. onThis specifictype dataof points.iterative Thissupervised processlearning is called active learning. UnlikeSince regularthe learninglearner wherechooses wethe needexamples, athe lotnumber of labeled examples, activeto learninglearn allowsa usconcept tocan learnoften withbe fewermuch exampleslower becausethan wethe picknumber whichrequired onesin tonormal learnsupervised fromlearning. HoweverWith this approach, there's is a challengerisk withthat thisthe approach.algorithm Weis mightoverwhelmed endby up with a bunch ofuninformative examples. that don'tRecent actuallydevelopments teachare usdedicated much.to Theremulti-label areactive newlearning,<ref ideasname="multi"/> inhybrid theactive fieldlearning<ref thatname="hybrid"/> focus onand active learning forin multiplea labelssingle-pass (on-line) context,<ref name="single-pass"/> combining different concepts from machinethe learningfield withof onlinemachine learning methods(e.g. They'reconflict alsoand lookingignorance) atwith hybridadaptive, approaches[[incremental thatlearning]] mixpolicies thesein ideasthe infield interestingof ways[[online machine learning]].

Large-scale active learning projects may benefit from [[crowdsourcing]] frameworks such as [[Amazon Mechanical Turk]] that include many [[human-in-the-loop|humans in the active learning loop]].