Skip to main content
Unpublished Paper
SampleRank: Training Factor Graphs with Atomic Gradients
  • Michael Wick
  • Khashayar Rohanimanesh
  • Kedar Bellare
  • Aron Culotta
  • Andrew McCallum, University of Massachusetts - Amherst
We present SampleRank, an alternative to contrastive divergence (CD) for estimating parameters in complex graphical models. SampleRank harnesses a user-provided loss function to distribute stochastic gradients across an MCMC chain. As a result, parameter updates can be computed between arbitrary MCMC states. SampleRank is not only faster than CD, but also achieves better accuracy in practice (up to 23% error reduction on noun-phrase coreference).
Publication Date
This is the pre-published version harvested from CIIR.
Citation Information
Michael Wick, Khashayar Rohanimanesh, Kedar Bellare, Aron Culotta, et al.. "SampleRank: Training Factor Graphs with Atomic Gradients" (2011)
Available at: