FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2189192010> ?p ?o ?g. }

• W2189192010 abstract "This thesis develops probabilistic programming as aproductive metaphor for understanding cognition, both with respectto mental representations and the manipulation of suchrepresentations. In the first half of the thesis, I demonstrate therepresentational power of probabilistic programs in the domains ofconcept learning and social reasoning. I provide examples of richlystructured concepts, defined in terms of systems of relations,subparts, and recursive embeddings, that are naturally expressed asprograms and show initial experimental evidence that they matchhuman generalization patterns. I then proceed to models ofreasoning about reasoning, a domain where the expressive power ofprobabilistic programs is necessary to formalize our intuitivedomain understanding due to the fact that, unlike previousformalisms, probabilistic programs allow conditioning to berepresented in a model, not just applied to a model. I illustratethis insight with programs that model nested in gametheory, artificial intelligence, and linguistics. In the secondhalf, I develop three inference algorithms with the dual intent ofshowing how to efficiently compute the marginal distributionsdefined by probabilistic programs, and providing building blocksfor process-level accounts of human cognition. First, I describe aDynamic Programming algorithm for computing the marginaldistribution of discrete probabilistic programs by compiling tosystems of equations and show that it can make inference in modelsof about reasoning tractable by merging and reusingsubcomputations. Second, I introduce the setting of amortizedinference and show how learning inverse models lets us leveragesamples generated by other inference algorithms to compileprobabilistic models into fast recognition functions. Third, Idevelop a generic approach to coarse-to-fine inference inprobabilistic programs and provide evidence that it can speed upinference in models with large state spaces that have appropriatehierarchical structure. Finally, I substantiate the claim thatprobabilistic programming is a productive metaphor by outlining newresearch questions that have been opened up by this line ofinvestigation." @default.
• W2189192010 created "2016-06-24" @default.
• W2189192010 creator A5076240321 @default.
• W2189192010 date "2015-01-01" @default.
• W2189192010 modified "2023-09-26" @default.
• W2189192010 title "Modeling Cognition with Probabilistic Programs: Representations and Algorithms" @default.
• W2189192010 cites W1499878068 @default.
• W2189192010 cites W1509562192 @default.
• W2189192010 cites W1525726177 @default.
• W2189192010 cites W1526328753 @default.
• W2189192010 cites W1528056001 @default.
• W2189192010 cites W1531283946 @default.
• W2189192010 cites W1533758202 @default.
• W2189192010 cites W1538748698 @default.
• W2189192010 cites W1561264340 @default.
• W2189192010 cites W1561981064 @default.
• W2189192010 cites W1562162659 @default.
• W2189192010 cites W1564759995 @default.
• W2189192010 cites W1575798196 @default.
• W2189192010 cites W1580018148 @default.
• W2189192010 cites W1589239973 @default.
• W2189192010 cites W1597739853 @default.
• W2189192010 cites W1606274310 @default.
• W2189192010 cites W1625390266 @default.
• W2189192010 cites W1729198099 @default.
• W2189192010 cites W1802388531 @default.
• W2189192010 cites W181022050 @default.
• W2189192010 cites W1877427040 @default.
• W2189192010 cites W1890754682 @default.
• W2189192010 cites W1892397232 @default.
• W2189192010 cites W1971964248 @default.
• W2189192010 cites W197196944 @default.
• W2189192010 cites W1987027398 @default.
• W2189192010 cites W1988914804 @default.
• W2189192010 cites W1993845689 @default.
• W2189192010 cites W1993979041 @default.
• W2189192010 cites W1998570995 @default.
• W2189192010 cites W1998754086 @default.
• W2189192010 cites W1998952369 @default.
• W2189192010 cites W2001566989 @default.
• W2189192010 cites W2002206192 @default.
• W2189192010 cites W2040908702 @default.
• W2189192010 cites W2047978125 @default.
• W2189192010 cites W2052810501 @default.
• W2189192010 cites W2055378604 @default.
• W2189192010 cites W2065412317 @default.
• W2189192010 cites W2068869385 @default.
• W2189192010 cites W2071544114 @default.
• W2189192010 cites W2081981374 @default.
• W2189192010 cites W2083719729 @default.
• W2189192010 cites W2089674328 @default.
• W2189192010 cites W2092919341 @default.
• W2189192010 cites W2098694378 @default.
• W2189192010 cites W2101714644 @default.
• W2189192010 cites W2104974755 @default.
• W2189192010 cites W2106425986 @default.
• W2189192010 cites W2107884096 @default.
• W2189192010 cites W2112578769 @default.
• W2189192010 cites W2115636120 @default.
• W2189192010 cites W2118382442 @default.
• W2189192010 cites W2119199958 @default.
• W2189192010 cites W2119766618 @default.
• W2189192010 cites W2123713131 @default.
• W2189192010 cites W2125781450 @default.
• W2189192010 cites W2129931127 @default.
• W2189192010 cites W2132089731 @default.
• W2189192010 cites W2133520183 @default.
• W2189192010 cites W2141038596 @default.
• W2189192010 cites W2142336306 @default.
• W2189192010 cites W2143516773 @default.
• W2189192010 cites W2148886952 @default.
• W2189192010 cites W2152096446 @default.
• W2189192010 cites W2155941727 @default.
• W2189192010 cites W2158522418 @default.
• W2189192010 cites W2158530728 @default.
• W2189192010 cites W2169898528 @default.
• W2189192010 cites W2185186942 @default.
• W2189192010 cites W2185558310 @default.
• W2189192010 cites W2197329920 @default.
• W2189192010 cites W2223465455 @default.
• W2189192010 cites W2396242484 @default.
• W2189192010 cites W2399803916 @default.
• W2189192010 cites W2402902442 @default.
• W2189192010 cites W2571600439 @default.
• W2189192010 cites W2623152124 @default.
• W2189192010 cites W2767589426 @default.
• W2189192010 cites W2768857716 @default.
• W2189192010 cites W2799095127 @default.
• W2189192010 cites W2911246871 @default.
• W2189192010 cites W2934944248 @default.
• W2189192010 cites W2963524296 @default.
• W2189192010 cites W2963704192 @default.
• W2189192010 cites W32064870 @default.
• W2189192010 cites W402698555 @default.
• W2189192010 cites W605810508 @default.
• W2189192010 cites W65465483 @default.
• W2189192010 cites W68436435 @default.
• W2189192010 cites W88990729 @default.
• W2189192010 hasPublicationYear "2015" @default.
• W2189192010 type Work @default.

• next