Home > MRFLearning > runJTgrid.m

Junction Tree Inference for a rectangular grid Boltzmann m/c Function that take a rectangular grid Boltzmann m/c (with 0/1 outputs) and runs a node clustering version of the general Junction Tree algorithm. Returns the node clusters and their final potentials (the marginals). Inputs: model: struct with fields numRows: # rows, R, in the grid numCols: # columns, C, in the grid alpha : (R X C) matrix of node biases wHor : (R X C-1) matrix of horizontal edge weights wVer : (R-1 X C) matrix of vertical edge weights Returns: sNodes: struct with fields nodes: matrix where row i gives the nodes belonging to supernode i. Size is ( number of supernodes X (R+1) ) pot : matrix where row i gives the marginal of supernode i. Size is ( number of supernodes X (2^(R+1))

- JTsampleGrid Exact sampling for rectangular grid boltzmann m/c's using junction tree.
- MLgrid Estimate the ML parameters for a rectangular grid boltzmann m/c,

Generated on Wed 28-Sep-2005 04:10:02 by