Learning and Applying Category Knowledge in Unsupervised Domains

Clapper, John P.; Bower, Gordon H.

doi:10.1016/s0079-7421(08)60121-8

Cited by 26 publications

(49 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Alternative views of unsupervised learning posit that human learners actively discover categories embedded within the stimulus set (J. R. Anderson, 1991;Clapper & Bower, 1991;Love, Medin, & Gureckis, in press). Category discovery models fare no better with respect to the present data.…”

Section: Discussionmentioning

confidence: 99%

The multifaceted nature of unsupervised category learning

Love

2003

Psychonomic Bulletin & Review

View full text Add to dashboard Cite

A substantial portion of category-learningresearch has focused on one learning mode-namely, classification learning (a supervised learning mode). Subsequently, theories of category learning have focused on how the abstract structure of categories (i.e., the co-occurrence patterns of feature values) affects acquisition. Recent work in supervised learning has shown that a learner's interactions with the stimulus set also plays an important role in acquisition. The present study extends this work to unsupervised learning situations involving simple one-dimensional stimuli. The results suggest that categorization performance is a function of both learning mode (i.e., study conditions) and learning problem (i.e., category structure). Unsupervised learning, like supervised learning, appears to be multifaceted, with different learning modes best paired with certain learning problems.

show abstract

Section: Discussionmentioning

confidence: 99%

The multifaceted nature of unsupervised category learning

Love

2003

Psychonomic Bulletin & Review

View full text Add to dashboard Cite

show abstract

“…Unsupervised recruitment in SUSTAIN bears a strong resemblance to recruitment in Adaptive Resonance Theory, Clapper and Bower's (1991) qualitative model, (Carpenter & Grossberg, 1987) and Hartigan's (1975) leader algorithm. When a new cluster is recruited (for both unsupervised and supervised learning situations) it is centered on the misclassified input pattern and the clusters' activations and outputs are recalculated.…”

Section: Mathematical Formulation Of Sustainmentioning

confidence: 99%

SUSTAIN: A Network Model of Category Learning.

Love

Medin²,

Gureckis

2004

Psychological Review

696

880

View full text Add to dashboard Cite

SUSTAIN (Supervised and Unsupervised STratified Adaptive IncrementalNetwork) is a model of how humans learn categories from examples. SUS-TAIN initially assumes a simple category structure. If simple solutions prove inadequate and SUSTAIN is confronted with a surprising event (e.g., it is told that a bat is a mammal instead of a bird), SUSTAIN recruits an additional cluster to represent the surprising event. Newly recruited clusters are available to explain future events and can themselves evolve into prototypes/attractors/rules. Importantly, SUSTAIN's discovery of category substructure is affected not only by the structure of the world, but by the nature of the learning task and the learner's goals. SUSTAIN successfully extends category learning models to studies of inference learning, unsupervised learning, category construction, and contexts where identification learning is faster than classification learning.

show abstract

“…Discovering new categories through unguided observation in the absence of explicit feedback or reinforcement is a fundamental ability that underlies much of human learning (see, e.g., Billman & Knutson, 1996;Clapper & Bower, 1991Mervis, 1987;Murphy, 2002;Nelson, 1974). This kind of self-directed pattern detection is often referred to as observational or unsupervised learning, in contrast to supervised learning, in which categories are defined in advance and participants receive corrective feedback from an external teacher (see, e.g., Bruner, Goodnow, & Austin, 1967;Medin & Schaffer, 1978;Posner & Keele, 1968, 1970Smith & Medin, 1981).…”

mentioning

confidence: 99%

When more is less: Negative exposure effects in unsupervised learning

Clapper

2006

Memory & Cognition

Self Cite

View full text Add to dashboard Cite

In this article, two broad classes of models of unsupervised learning are compared: correlation tracking models, according to which learning is expected to increase monotonically with exposure to instances, and category invention models, which can accommodate specific violations of monotonicity (negative exposure effects). In two experiments, increasing the number of training instances had a negative rather than a positive effect on unsupervised learning, a clear violation of monotonicity. The results of these experiments are then compared with the predictions of two computational models, one a category invention model and the other a correlation tracking model. The category invention model was able to reproduce the qualitative pattern of results from the human data, whereas the correlation tracking model was not. Overall, these results provide strong evidence for the existence of a discrete category invention process in unsupervised learning.

show abstract

Learning and Applying Category Knowledge in Unsupervised Domains

Cited by 26 publications

References 19 publications

The multifaceted nature of unsupervised category learning

The multifaceted nature of unsupervised category learning

SUSTAIN: A Network Model of Category Learning.

When more is less: Negative exposure effects in unsupervised learning

Contact Info

Product

Resources

About