Yet More Efficient EM Learning for Parameterized Logic Programs by Inter-Goal Sharing

In previous research, we presented a dynamicprogramming-based EM (expectation-maximization) algorithm for parameterized logic programs, which is based on the structure sharing with tabled search. It is also shown that this general framework achieves the same time complexity as that of the specialized algorithms. The efficiency is brought by sharing common (partial) paths in the derivation tree for a given goal, but such sharing is incomplete in the sense that it is not allowed to share the paths appearing in different derivation trees. In this paper, we introduce a general idea called ‘inter-goal sharing’ where different goals can share the common derivation paths. Inter-goal sharing achieves the full sharing of derivation paths and hence makes EM learning more compact and efficient in practical cases. Experimental results with both artificial and real linguistic data show that the proposed method runs 2-6 times more compactly and faster than the previous approach.