Lifted Model Construction under Approximate Commutativity

Lifted inference algorithms scale probabilistic inference by exploiting indistinguishable objects, but real‑world data often yields only approximately commutative factors. The paper introduces ε‑commutativity, a relaxation that tolerates small deviations from exact invariance, and shows how it can be used to build lifted models and perform inference with provable error bounds. Empirical results confirm that queries remain accurate while runtime is reduced.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.