Learning overhypotheses with hierarchical Bayesian models
Learning overhypotheses with hierarchical Bayesian models
复制标题
DOI：
10.1111/j.1467-7687.2007.00585.x
复制
发表时间：
2007-05-01
影响因子：
3.7
通讯作者：
Tenenbaum, Joshua B.
中科院分区：
文献类型：
--
作者：
Kemp, Charles;Perfors, Amy;Tenenbaum, Joshua B.
归纳学习是不可能的，没有过度假设，或对学习者所考虑的假设的限制。其中一些过度假设必须是先天的，但我们认为，分层贝叶斯模型可以帮助解释如何获得其余的。为了说明这一点，我们开发了两种超假设的模型-关于特征变异性的超假设（例如，单词学习中的形状偏差）和关于将类别分组为对象和物质等本体类的超假设。
Inductive learning is impossible without overhypotheses, or constraints on the hypotheses considered by the learner. Some of these overhypotheses must be innate, but we suggest that hierarchical Bayesian models can help to explain how the rest are acquired. To illustrate this claim, we develop models that acquire two kinds of overhypotheses - overhypotheses about feature variability (e.g. the shape bias in word learning) and overhypotheses about the grouping of categories into ontological kinds like objects and substances.