AI · Theory

Constraint Propagation

Theory

153/215

Constraint Propagation

Constraint Propagation is about choosing a sequence of actions toward a goal, sometimes with other agents or chance. State the goal, the actions, and whether the world is deterministic. Planning ≠ classification.

Output — idea / do / check for Constraint Propagation. Fill those three in the viva.

Tiny Constraint Propagation story — get the key, open the door, reach the gold. That beats a UML diagram.

Diagram

data → train / test
           │
           ▼
         model
           │
           ▼
        predict

Exam tip

Goal + actions + deterministic or not.

Example

# Constraint Propagation
print("idea :", "Constraint Propagation")
print("do   : one tiny example on paper")
print("check: one failure case")

Output: idea / do / check for Constraint Propagation. Fill those three in the viva.

Short notes

DefConstraint Propagation — actions toward a goal.
RuleState the world type.
TrapConstraint Propagation — calling every ML model “planning”.

AI · Theory

Constraint Propagation

← All stacks

Theory

153/215

Constraint Propagation

Output — idea / do / check for Constraint Propagation. Fill those three in the viva.

Tiny Constraint Propagation story — get the key, open the door, reach the gold. That beats a UML diagram.

Diagram

data → train / test
           │
           ▼
         model
           │
           ▼
        predict

Exam tip

Goal + actions + deterministic or not.

Example

# Constraint Propagation
print("idea :", "Constraint Propagation")
print("do   : one tiny example on paper")
print("check: one failure case")

Output: idea / do / check for Constraint Propagation. Fill those three in the viva.

Short notes

DefConstraint Propagation — actions toward a goal.
RuleState the world type.
TrapConstraint Propagation — calling every ML model “planning”.

Constraint Propagation

Example

Short notes

Questions

Constraint Propagation

Example

Short notes

Questions