AI-Hackathon: ELO competition

In the style of hackathons, demo coder parties and game jams, an AI competition will be held. Over a period of two weeks, participants will have the opportunity to acquire practical AI expertise in a fun way. They will develop their own game AIs for board games, which are currently the source of much of the fascination with modern AI (see Alpha Zero). Each event concludes with a competition for the AIs submitted, in which an ELO-score (based on the FIDE chess association) is determined and certified.

Participants will be provided with a range of training materials to help them acquire the necessary AI knowledge.

On the one hand, the screencasts provide an overview of the details of the hackathon, and on the other hand, they offer detailed descriptions of AI implementations, particularly with regard to the board games on offer. In addition, the necessary knowledge for using the competition framework and finally an excursus on the topic of monetization are covered.

 

 

Hackathon and Basic-Strategies

Strategies 01
  •  Shannon A
  •  Evaluation Functions
  •  Implementations of Shannon A (Minimax & Negamax)

Watch German Video

Strategies 02
  •  Speed-Up Techniques
  •  Shannon B
  •  Implementation of Shannon B
  •  Transposition-Tables & Zobrist Hashing

Watch German Video

Strategies 03
  • Boardgame AIs
  • Disk-Square Approach
  • Mobility-Based Algorithms
  • Pattern Recognition

Watch german video

Framework and Environment
  •  The Gameboard Class
  •  The Elo-Arena
  •  Example: Random Decider
  •  The Gui
  •  Submission procedure

Watch German video

Monetization of Video Games
  • Motivation and Market Overview
  • Basics of Economies, Game-development and Gambling
  • The Gamer, its Motivation and Behaviour
  • Monetization models
  • Dark Patterns
  • Ai for Monetization in Video Games

Watch German Video

Advanced Topics

General Ideas of Reinforcement Learning
  • What is Reinforcement Learning
  • Markov-Chains
  • Episodic vs Continuous Task
  • Value, Q-Value, Advantage, Regret
  • Bellmann Equation
  • Dynamic Programming
  • On Policy vs Off Policy Learning
  • Exploration vs Exploitation Dilemma
  • K-arm bandit and UCB strategies

Watch German Video

PPO and SAC
  • PPO (Proximal Policy Optimization)
  • SAC (Soft Actor Critic)

Watch German Video

Alpha Zero
  • History / Overview
  • Monte Carlo Tree Search
  • Algorithm: Alpha Zero
  • Implementation

Watch German Video

Prof. Dr. Christoph Lürig
Prof. Dr. Christoph Lürig
Professor FB Informatik

Contact

+49 651 8103-372

Location

Schneidershof | Building X | Room 8
Fabian Fell
Beschäftigter FB Informatik

Location

Schneidershof | Building O | Room 5
back-to-top nach oben