SoPU: Sums of Products United Research ClubVision
Non-standard methods to evaluate large and structured sums of productsespecially ones involving moderately exponential algorithmshave great prospects to significantly advance the state of the art in algorithm theory and computational statistics.
SoPU's current mission is to implement the vision in a prototyping manner by studying
- algorithm theory of computing sums of products,
- sums of products in computational statistics,
- applications in science and technology.
- Fast sum-product algorithms for set functions
- Efficient algorithms for structure learning in Bayesian networks
- Redescription mining: models, heuristics, and exact algorithms
- Bounding by products: global bounds via local views
- Some ambitious, next-phase ideas are at proposal stage, pending funding.
- Esther Galbrun, doctoral student (co-advisor Hannu Toivonen)
- Kustaa Kangas, doctoral student
- Mikko Koivisto, advisor, coach
- Janne Korhonen, doctoral student (co-advisor Petteri Kaski)
- Teppo Niinimaki, doctoral student
- Pekka Parviainen, PhD March 2012, since June 2012 a postdoc researcher at KTH, Sweden.
The research is funded by scattered small to moderate-size grants for projects that loosely fit SoPU's mission: by the Academy of Finland (M.K. 8/2008-10/2013), the doctoral programmes FICS (P.P. 1/2010-6/2011) and Hecse (J.K. 1/2010-12/2013; T.N. 1/2011-12/2014), the NIH (UCLA subaward, PI: Heikki Mannila, 7/2007-6/2012), the Algodan CoE, and HIIT.
Last modified Feb 11, 2013.