Lecture 6

To show is np hard it suces to nd a transformation t

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: time using a non-deterministic oracle (i.e., you can verify its guesses in polytime). • Thm: 3-SAT is NP-complete. • To show Π is NP-hard, it suffices to find a transformation T ∈ P from another NP-hard problem Π (e.g., 3-SAT) since • Defn: Π is NP-hard if ∀Π ∈ N P. ∃T ∈ P. Π → Π. • Defn: Π is NP-complete if it is NP-hard and in NP. • To show Π is NP-complete, show it is NP-hard and that you can check (oracular) guesses in poly-time. • Conjecture: P = N P NP−hard NP NP−hard NP P P NP complete Exact inference in discrete Bayes nets is NP-complete • Thm: the decision problem “Is PB (Xi = x) > 0?” is NP-complete. • Proof. To show in NP: Given an assignment X1:n, we can check if Xi = x and then check if P (X1:n) > 0 in poly-time. To show NP-hard: we can encode any 3SAT problem as a polynomially sized Bayes net, as shown below. • P (X = 1|q1:n) > 0 iff q1:n is a satisfying assignment. 4 ¥ ¤ 4 ¢ 4 £ 4 ¡ 4 T T NP → Π → Π T NP complete Complexity of approximate inference • Defn: An estimate ρ has absolute error |P (y |e) − ρ| ≤ . for P (y |e) if • Defn: An estimate ρ has relative error for P (y |e) if ρ ≤ P (y |e) ≤ ρ(1 + ) 1+ • Thm: Computing P (Xi = x) with re...
View Full Document

This document was uploaded on 03/28/2014 for the course CS 532 at UBC.

Ask a homework question - tutors are online