Efficient Weight Learning for Markov Logic Networks
Efficient Weight Learning for Markov Logic Networks
Daniel Lowd,Pedro M. Domingos
2007 · DOI: 10.1007/978-3-540-74976-9_21
European Conference on Principles of Data Mining and Knowledge Discovery · 243 Citations
TLDR
This paper focuses on two approaches that avoid computing the partition function: diagonal Newton and scaled conjugate gradient, and achieves order-of-magnitude speedups, or more accurate models given comparable learning times.