A Unifying Computational Framework for Teaching and Active Learning. (13th January 2019)

• Record Type: Journal Article
• Authors: Yang, Scott Cheng‐Hsin; Vong, Wai Keen; Yu, Yue; Shafto, Patrick
• Is Part Of: Topics in cognitive science. Volume 11:Number 2(2019)
• Page Start: 316
• Page End: 337
• Publication Date: 2019-01-13
• Subjects: Bayesian model -- Active learning -- Bayesian teaching; Cognitive science -- Periodicals; Cognitive Science -- Periodicals; 153.05
• DOI: 10.1111/tops.12405
• Languages: English
• ISSNs: 1756-8757

Abstract

Abstract: Traditionally, learning has been modeled as passively obtaining information or actively exploring the environment. Recent research has introduced models of learning from teachers that involve reasoning about why they have selected particular evidence. We introduce a computational framework that takes a critical step toward unifying active learning and teaching by recognizing that meta‐reasoning underlying reasoning about others can be applied to reasoning about oneself. The resulting Self‐Teaching model captures much of the behavior of information‐gain‐based active learning with elements of hypothesis‐testing‐based active learning and can thus be considered as a formalization of active learning within the broader teaching framework. We present simulation experiments that characterize the behavior of the model within three simple and well‐investigated learning problems. We conclude by discussing such theory‐of‐mind‐based learning in the context of core cognition and cognitive development. Abstract : According to rational pedagogy models, learners take into account the way in which teachers generate evidence, and teachers take into account the way in which learners assimilate that evidence. The authors develop a framework for integrating rational pedagogy into models of active exploration, in which agents can take actions to influence the evidence they gather from the environment. The key idea is that a single agent can be both teacher and learner.