Learning overhypotheses with hierarchical Bayesian models.

69 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Probabilistic models of cognition: exploring representations and inductive biases. (2010, unknown): “learn” words [41], as well as categories [42] and causal relationships [28, 43], by”
Reconstructing constructivism: Causal models, Bayesian learning mechanisms and the theory theory (2012, unknown): “Kemp et al. (2007) demonstrated how overhypotheses can be learned in hierarchical Bayesian models, and building on these ideas, Goodman et al. (2011) have used hierarchical Bayesian modeling to provide a striking set of computational results they call “the blessing of abstraction” (p. 5).”
One-shot learning of generative speech concepts (2014, background): “The lesioned models and Dynamic Time Warp demonstrated inferior performance on the classiﬁcation and generation tasks, adding to previous evidence that compo-sitionality and learning-to-learn are important for one-shot learning (Kemp et al., 2007; Salakhutdinov et al., 2012; Lake et al., 2013).”
The Surfeit of the Stimulus: Analytic Biases Filter Lexical Statistics in Turkish Laryngeal Alternations (2011, background): “The same-feature constraint on vowel-consonant interactions is thus an " overhypothesis " in the sense of Goodman (1955) and Kemp et al. (2007): a meta-level hypotheses that constrains the form of possible specific hypotheses and generalizations induced from the data.”