DeepSWIP introduces a single-world counterfactual semantics for DeepProbLog, enabling causal reasoning through neural materialization and weighted model counting. It achieves exact inference under finite grounding and unique-supported-model assumptions, with experiments showing a 2.14× speedup and improved calibration over DeepTwin and AIPW estimators.
DeepSWIP: Counterfactual Reasoning in Neural Probabilistic Logic
from English