Please enable JavaScript.
Coggle requires JavaScript to display documents.
Adversarial Search, A form of search used when there is an “opponent” —…
Adversarial Search
-
-
-
AND-OR Search Trees
OR nodes
- Represent a choice made by the agent
- “What move should I take?”
AND nodes
- Represent all possible responses by the opponent or nature
- “What could happen next, no matter what?”
-
Example
- In a 2-player game, choosing to attack (OR) leads to all the opponent’s counterattacks (AND) — the agent must be ready for all responses
Minimax Algorithm
-
-
-
→ Example:
- In Tic Tac Toe, you evaluate your move and all the opponent’s responses to pick the safest winning path
-
Alpha-Beta Pruning
-
-
If branch can't improve result, it gets “pruned” (ignored)
Saves time and memory, especially in deep trees
→ Example:
- You don't need to explore a losing move if you've already found a better one
-
Real-World Applications
Game AI
- Chess engines, boss fights, fighting game AI
Cybersecurity
- Simulate attacks/defense between attacker and system
Negotiation bots
- Predict and counter opposing offers
Example:
- AlphaZero beat human chess grandmasters by learning to predict every possible move
-
-
- A form of search used when there is an “opponent” — another agent trying to prevent you from achieving your goal.
- In standard pathfinding or goal-based search, the environment is passive. In adversarial search, it fights back.
- A strategy where one agent (MAX) tries to maximize score, while the opponent (MIN) tries to minimize it.
- When more than two agents play against or around each other.
-
