state-space-repr-intro.wiki (1083B)
1 = State Space Representations Intro = 2 * real world is complex, state space must be _abstracted_ for problem solving 3 * abstract states map to sets of real states 4 * abstract action maps to combination of real actions 5 * abstract solution = set of real paths that are solutions in real world 6 * there may be multiple different state space representations 7 8 Problem-solving agent uses this representation: 9 1. what are actions to move between states? 10 2. what are appropriate states & initial states? 11 3. what is the cost of an action? 12 4. goal: what are the successful world states? 13 5. search: determine possible sequences of actions leading to goal, choose 'best' sequence 14 6. execute: give solution, perform actions 15 16 State space representation (example - vacuum cleaner): 17 * start with real-life problem 18 * formulate abstract problem (states, actions) 19 * formulate concrete (clean house) & algorithmic goal (be in state 7 and 8) 20 * find solution (sequence of actions to get to state 7 or 8) 21 * execute plan (clean house according to abstract solution) 22