We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

stat.AP

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Statistics > Applications

Title: Simple trees in complex forests: Growing Take The Best by Approximate Bayesian Computation

Abstract: How can heuristic strategies emerge from smaller building blocks? We propose Approximate Bayesian Computation as a computational solution to this problem. As a first proof of concept, we demonstrate how a heuristic decision strategy such as Take The Best (TTB) can be learned from smaller, probabilistically updated building blocks. Based on a self-reinforcing sampling scheme, different building blocks are combined and, over time, tree-like non-compensatory heuristics emerge. This new algorithm, coined Approximately Bayesian Computed Take The Best (ABC-TTB), is able to recover a data set that was generated by TTB, leads to sensible inferences about cue importance and cue directions, can outperform traditional TTB, and allows to trade-off performance and computational effort explicitly.
Subjects: Applications (stat.AP)
Cite as: arXiv:1605.01598 [stat.AP]
  (or arXiv:1605.01598v2 [stat.AP] for this version)

Submission history

From: Eric Schulz [view email]
[v1] Thu, 5 May 2016 13:59:37 GMT (176kb,D)
[v2] Sat, 14 May 2016 14:41:13 GMT (223kb,D)

Link back to: arXiv, form interface, contact.