Human Semi‐Supervised Learning

Abstract

Most empirical work in human categorization has studied learning in either fully supervised or fully unsupervised scenarios. Most real-world learning scenarios, however, are semi-supervised: Learners receive a great deal of unlabeled information from the world, coupled with occasional experiences in which items are directly labeled by a knowledgeable source. A large body of work in machine learning has investigated how learning can exploit both labeled and unlabeled data provided to a learner. Using equivalences between models found in human categorization and machine learning research, we explain how these semi-supervised techniques can be applied to human learning. A series of experiments are described which show that semi-supervised learning models prove useful for explaining human behavior when exposed to both labeled and unlabeled data. We then discuss some machine learning models that do not have familiar human categorization counterparts. Finally, we discuss some challenges yet to be addressed in the use of semisupervised models for modeling human categorization.Keywords: Category learning; Semi-supervised learning; Machine learning Cognitive psychology has long had an interest in understanding human categorization: how we come to conceive of objects in the world as belonging to different categories, and how we use categories to draw inferences about the unobserved properties of objects. Toward this end, one of the most commonly used experimental paradigms has been supervised category learning: On each trial, the participant views a stimulus and must guess to which of a small number of categories it belongs. Feedback is provided that indicates either whether the guess was correct or what the correct answer was-the learning is supervised in this sense. The experimenter then measures how rapidly the participant learns to generate correct inferences about category membership, and how the acquired knowledge generalizes to novel stimuli.Correspondence should be sent to Bryan R. Gibson,

A little labeling goes a long way: Semi‐supervised learning in infancy

“…Here, we have shown that infants do incorporate unlabeled exemplars after a few labeled exemplars have been presented. What remains to be seen is whether and how these unlabeled exemplars prompt a shift in the content of the categories, as they do in adult and machine learners (Gibson et al., ). By examining how infants' categories change to reflect each new exemplar, we can also begin to more precisely model infants' learning process.…”

Robust speech perception: Recognize the familiar, generalize to the similar, and adapt to the novel.

“…We have also tentatively proposed that there is a parallel trade-off at a higher level, in estimating the distribution of generative models themselves across situations. This suggests that listeners might be able to use exemplars in generative model parameter space (rather than acoustic space) to estimate, in an approximate, boundedly-rational way, the overall distribution of generative models across situations (Ashby & Alfonso-Reese, 1995; Gibson et al, 2013; Griffiths, Sanborn, Canini, & Navarro, 2008) and to adapt to the current situation (Shi et al, 2010). However, there is also a benefit from abstracting away from individual situations by explicitly representing summaries of different group-level distributions in order to generalize across groups.…”

“…M. Munson, 2011). Further work is required to better understand the relative role of supervised and unsupervised learning (or the continuum between these extremes; e.g., Gibson et al, 2013; Zhu, Rogers, Qian, & Kalish, 2007) both during acquisition and in speech perception in adults.…”

Arrangement and Applying of Movement Patterns in the Cerebellum Based on Semi-supervised Learning

“…This new learning strategy is potentially a quantitative tool to understand human category learning, where most of the input is self-evidently unlabeled [13]. Much of human concept learning involves a small amount of direct instruction (e.g., parental labeling of objects during childhood) combined with large amounts of unlabeled experience (e.g., observation of objects without naming or counting them, or at least without feedback) [5,[14][15][16][17].…”