4 - CSP Flashcards
Constrained Satisfaction Problems (CSPs) use ________.
Constrained Satisfaction Problems (CSPs) use FACTORED REPRESENTATION OF EACH STATE.
Positive Goal Test
Variable satisfies all constraints
Unary Constraint + Example
Involves SINGLE variable (SA != green)
Binary Constraint + Example
Involves PAIRS of variables (SA != WA)
Higher-Order Constraint + Example
Involves 3+ variables (SA != WA && WA != NT)
Higher-Order constraints can be rewritten as ________.
… BINARY CONSTRAINTS
Backtracking Search
Assign all possible values to node and continue with next node. (DFS for CSPs)
Minimum Remaining Value Heuristic
Choose the variable with the fewest possible values first (-> prune search tree)
Arc-Consistency
Every value x_i in D_i has a counterpart x_j in D_j that satisfies the binary constraint (Ex.: Y = X^2)
Least Constraining Value Heuristic
Select value that maximizes the remaining possible values for the next nodes (-> select first value)
Degree Heuristic
Assign variable involved with most constraints on other unassigned values first
Inference
Reduce assignable domain values based on logical/trivial conclusions
Tree-structured CSPs can be solved in ________ iff ________.
Tree-structured CSPs can be solved in O(nd^2) TIME iff the CONSTRAINT GRAPH HAS NO LOOPS.
Tree Decomposition for nearly tree-structured CSPs
Decomposing the CSP into a set of connected subproblems
Conditioning for nearly tree-structured CSPs
Instantiate a variable and prune its neighbors’ domains