A Comparison of Time-Space Schemes for Graphical Models

Robert Mateescu, Rina Dechter

We investigate three parameterized algorithmic schemes for graphical models that can accommodate trade-offs between time and space: 1) AND/OR Adaptive Caching (AOC(i)); 2) Variable Elimination and Conditioning (VEC(i)); and 3) Tree Decomposition with Conditioning (TDC(i)). We show that AOC(i) is better than the vanilla versions of both VEC(i) and TDC(i), and use the guiding principles of AOC(i)to improve the other two schemes. Finally, we show that the improved versions of VEC(i) and TDCI can be simulated by AOC(i), which emphasizes the unifying power of the AND/OR framework.

URL: http://www.ics.uci.edu/~mateescu