Scribd
Upload
84 views2 pages

What Is A Key Point of General Game Playing?

A key point of general game playing is the capability to effectively play various kinds of games without human interference by heuristically learning from the formal game description. To develop a general game playing system that can win tournaments, one must first understand the game description and then perform automatic reasoning during gameplay to learn how to win. This involves feature extraction, search algorithms, and efficient learning algorithms. Deep learning techniques can help with feature extraction and Monte Carlo tree search is a promising technique for heuristic search. With continued advances in machine learning like deep neural networks, it may be possible to develop a world-champion level computer program for most games.

Uploaded by

Văn Thuấn Ngô
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
84 views2 pages

What Is A Key Point of General Game Playing?

A key point of general game playing is the capability to effectively play various kinds of games without human interference by heuristically learning from the formal game description. To develop a general game playing system that can win tournaments, one must first understand the game description and then perform automatic reasoning during gameplay to learn how to win. This involves feature extraction, search algorithms, and efficient learning algorithms. Deep learning techniques can help with feature extraction and Monte Carlo tree search is a promising technique for heuristic search. With continued advances in machine learning like deep neural networks, it may be possible to develop a world-champion level computer program for most games.

Uploaded by

Văn Thuấn Ngô
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Question 9.

What is a key point of general game playing?

A key point of general game playing is the capability to effectively play various kinds of game with
requirement of the formal description of the game and without interfering by human during game
playing and predefining any solving algorithm. Therefore, the GGP have to itself heuristically learn
from the game description (perhaps also from its rivals) to automatically reason for the best move in
a game.
How can you develop a general game playing system to win the tournament?

The first task involves understanding about game description (it may be expresses in logical rules,
hence easier to achieve). The subsequent step relates to automatic reasoning during game runtime. It
is the period of time for GGP to learn how to win in a game. During this process, it emerges three
basic problems including feature extraction and combinatorial techniques to improve learning

efficiency, search/planning algorithms independent of any domain knowledge, efficient


learning algorithms that can learn new environments. To extract features and study
combinatorial mind game, we can take advantage of deep learning techniques and the
automatic feature extraction of discrete input. To perform heuristic search or planning, we
can use Monte-Carlo Tree Search (MCTS), which is currently one of the most promising
techniques in GGP. To adapt with a new environment, we can consult more than one
strategy for solving the game to suggest the best fit strategy to the game.
Is it possible to develop a world-champion level computer program by machine learning
approach? If so, how?

In my opinion, designing a computer-world-game champion which bases on machine-learning


approaches is absolutely feasible. Today, it might be impossible but in the future, it may be possible.
With current achievement of machine learning techniques, a computer is initially able to imitate the
human way of thinking. At this time, if we only bases on the current ability of machine learning,
developing a computer game champion is a little tricky. However, I believe that by using deep neural
network (a state-of-the-art machine learning technique) for state evaluation in GGP, we may have a
chance to create a powerful computer player. In particularly, the study Neural Networks for State
Evaluation in General Game playing by Daniel Michulke and Michael Thielscher showed that the
winning rate for the computer is up to 70%. For further investigation on GGP with Deep neural
network, it is thought that a computer grand master for most of the game could be constructed.

References:
1. General Game Playing: Overview of the AAAI Competition - Michael Genesereth,
Nathaniel Love, and Barney Pell
2. Recent Advances in General Game Playing - Maciej wiechowski, HyunSoo Park, Jacek
Madziuk, and Kyung-Joong Kim

3. CI in General Game Playing - to date achievements and perspectives - Karol Waledzik and
Jacek Mandziuk
4. Self-Adaptation of Playing Strategies in General Game Playing -Maciej Swiechowski and
Jacek Ma ndziuk, Senior Member, IEEE
5. Neural Networks for State Evaluation in General Game Playing - Daniel Michulke and
Michael Thielscher, Department of Computer Science Dresden University of Technology
6. Intelligent novice and GGP - Hiroyuki Iidas Lecture Note - Game Informatics

You might also like