BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:The discovery of structural form - Dr Per Ola Kristensson (Univers
 ity of Cambridge)
DTSTART:20081027T111100Z
DTEND:20081027T121500Z
UID:TALK14232@talks.cam.ac.uk
CONTACT:Philip Sterne
DESCRIPTION:By Charles Kemp\, and Joshua B. Tenenbaum\n\nAbstract below:\n
 \nAlgorithms for finding structure in data have become increasingly import
 ant both as tools for scientific data analysis and as models of human lear
 ning\, yet they suffer from a critical limitation. Scientists discover qua
 litatively new forms of structure in observed data: For instance\, Linnaeu
 s recognized the hierarchical organization of biological species\, and Men
 deleev recognized the periodic structure of the chemical elements. Analogo
 us insights play a pivotal role in cognitive development: Children discove
 r that object category labels can be organized into hierarchies\, friendsh
 ip networks are organized into cliques\, and comparative relations (e.g.\,
  “bigger than” or “better than”) respect a transitive order. Stand
 ard algorithms\, however\, can only learn structures of a single form that
  must be specified in advance: For instance\, algorithms for hierarchical 
 clustering create tree structures\, whereas algorithms for dimensionality-
 reduction create low-dimensional spaces. Here\, we present a computational
  model that learns structures of many different forms and that discovers w
 hich form is best for a given dataset. The model makes probabilistic infer
 ences over a space of graph grammars representing trees\, linear orders\, 
 multidimensional spaces\, rings\, dominance hierarchies\, cliques\, and ot
 her forms and successfully discovers the underlying structure of a variety
  of physical\, biological\, and social domains. Our approach brings struct
 ure learning methods closer to human abilities and may lead to a deeper co
 mputational understanding of cognitive development
LOCATION:TCM Seminar Room\, Cavendish Laboratory\, Department of Physics
END:VEVENT
END:VCALENDAR