100% Guaranteed Results


MAS – Multi-Agent Systems: Final Project Mafia game simulation Solved
$ 20.99
Category:

Description

5/5 – (1 vote)

Minh Ngo Georgi Terziev
10897402 11128739
Arthur Braˇzinskas
1 Introduction
In this project we are going to simulate the Mafia game, also known as Werewolf. The research question in this case will be to determine a required ratio of mafiosi to innocents in the game to make the game balanced. Previous research has been conducted on a very limited amount of roles (mafia against townfolk). In our project, player roles will be expanded to include special roles, such as investigators, doctors, and lovers, which have different abilities. We will also introduce a set of properties that determine players as careless or mistrustful, and experiment with different player strategies.
2 Mafia game
3 Multi-agent system formulation
We continue by further discussing the hyperparameters of the simulation, and scetching a draft for the agents’ beliefs, desires, intentions and strategies.
4 Simulation hyperparameters
• Amount of players (at least three)
• Amount of people for each role:
1. At least one mafioso
2. At most one policeman
3. At most one doctor
4. At most one lover
5. At least two innocents
• Level of mistrustfulness determines the probability that one player can convince another that he is not a mafioso.
5 Beliefs
Each agent has beliefs about the roles of the other agents. At the beginning of the game, beliefs are initialized as follows:
• Mafiosi know each others identities.
• Innocents only know their own roles.
Beliefs will be updated after each round of the game, dependent on the player’s observations. For example, players can change their beliefs after the day’s vote. Additionally, the policeman updates his beliefs depending on his checks. If the policeman decides to share some of his beliefs with the other players, then each player can decide on their own wether to believe that he is the policeman or not, and concequently update their beliefs.
6 Desires
Desires are structured depending on the goal of the particular agent:
• For mafiosi: ”Kill innocents”, ”Hide” (dependent on the chance to be observed).
• For innocents and the policeman: ”Find mafia”
• For the doctor: ”Treat innocents”
• For the lover: ”Sleep with mafia”
7 Intentions
Based on desires we derive the following intentions for each role:
• For mafiosi: ”kill person A”, ”convince person B”
• For innocents: ”vote against person A”, ”convince person B”
• For the doctor: ”treat person A”, ”vote against person A”, ”convince person B”
• For the lover: ”Sleep with person A”, ”vote against person A”, ”convince person B”
8 Strategies
• To kill random person that is not mafia.
• To kill person that has a high chance to observe mafia.
The policeman can choose:
• To never share his insights
• To share immediately his insights (but then the mafia will know who he is).
• To share information when at least one mafioso has been killed.
• a simple majority vote – trust the most trusted people.
• a highest commulative reward strategy – trust poeple with high rate of success so far.
The proposed strategies can be evaluated to determine their efficiency in the Mafia game.
References
Wikipedia. Mafia (party game). URL https://en.wikipedia.org/wiki/Mafia_(party_ game).

Reviews

There are no reviews yet.

Be the first to review “MAS – Multi-Agent Systems: Final Project Mafia game simulation Solved”

Your email address will not be published. Required fields are marked *

Related products