7 Wonders Agent: A Proposal

7 Wonders Agent:A Proposal

Computer Science Thesis ProposalFall 2012

Ben GardinerAdvisor: Prof. Haiyan Cheng

Outline

I. Introduction and MotivationII. BackgroundIII. Rules for 7 WondersIV. Project ProposalV. Timeline

Introduction

• Look to inspiration from previously taken classes.

• Machine Learning class focused on artificial intelligence, through the game Go.

• Games other than Go are more interesting to me.

• Goal: Combine love of games with senior research project.

Motivation

• Use model to represent decisions.• Gameplay provides an ideal model for

decisions.• Creating an AI that plays a game is simpler

than creating an AI for a real-world task.

Background

Euro-Games

7 Wonders Rules

7 Wonders : a Wonder

Modified Rules Proposal

• Have 3 “Generations” of modified rules.• Write the agent for the simplest rule set.• Augment the agent for each more

complicated rule set.• Use the Monte-Carlo tree search approach.

Simplified Rules

• Ignore the “Wonder”• Players cannot “Buy” resources from

neighbors

Simplified Rules 2

• Purchase Resources from Neighbors

Final (Full) Rules

• Wonder can be built• Cards can be built for free

TimetableDate Deliverable

14-Jan First Day of Classes

31-Jan Initial game rules implemented

14-Feb Agent for initial rules completed

28-Feb Second iteration of rules implemented

14-Mar Second Agent Implemented

28-Mar Third Generation Rules

11-Apr Third Agent Implemented

17-Apr SSRD

5-May Final Deadline