Summary of Learning with Generalised Card Representations For “magic: the Gathering”, by Timo Bertram et al.

Learning With Generalised Card Representations for “Magic: The Gathering”

by Timo Bertram, Johannes Fürnkranz, Martin Müller

First submitted to arxiv on: 8 Jul 2024

GrooveSquid.com Paper Summaries

GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!

Summary difficulty	Written by	Summary
High	Paper authors	High Difficulty Summary Read the original abstract here
Medium	GrooveSquid.com (original content)	Medium Difficulty Summary The researchers tackle the challenge of building decks in collectable card games like “Magic: The Gathering” using AI models. They focus on developing generalizable representations for cards that can be used to predict human deck-building decisions. The team explores different types of features, including numerical, nominal, and text-based attributes, as well as meta information from third-party services. The results show that the choice of representation has little impact on learning to predict card selections within known sets but can significantly improve performance on unseen cards. The model is able to accurately predict human choices for 55% of completely new cards, indicating a deep understanding of card quality and strategy.
Low	GrooveSquid.com (original content)	Low Difficulty Summary Artificial intelligence helps people build better decks in collectible card games like Magic: The Gathering. Building decks is tricky because there are many different cards and they have complex meanings. New cards are released all the time, which makes it hard to predict what will work well together. Most AI models for building decks only work with a fixed set of cards, but this can be limiting in real-life situations. Researchers investigated ways to represent cards that would allow their AI model to work with new cards it had never seen before. They found that different types of information about the cards, such as numbers, words, and images, could help the model make better predictions.