RECORD
Improving Search with Supervised Learning in Trick-Based Card Games
Artificial Intelligence · Submitted 22 Mar 2019
abstract compiled · 2026-09-23 · no language-model review
The authors' main claim This technique, used in conjunction with Perfect Information Monte Carlo (PIMC) search, provides a substantial increase in cardplay strength in the popular trick-taking card game of Skat.
The compiled record
The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
Open the compiled record
The abstract, sentence by sentence
In trick-taking card games, a two-step process of state sampling and evaluation is widely used to approximate move values. While the evaluation component is vital, the accuracy of move value estimates is also fundamentally linked to how well the sampling distribution corresponds the true distribution. Despite this, recent work in trick-taking card game AI has mainly focused on improving evaluation algorithms with limited work on improving sampling. In this paper, we focus on the effect of sampling on the strength of a player and propose a novel method of sampling more realistic states given move history. In particular, we use predictions about locations of individual cards made by a deep neural network --- trained on data from human gameplay - in order to sample likely worlds for evaluation. This technique, used in conjunction with Perfect Information Monte Carlo (PIMC) search, provides a substantial increase in cardplay strength in the popular trick-taking card game of Skat.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | In trick-taking card games, a two-step process of state sampling and evaluation is widely used to approximate move values. | background 1.00 | 0.05 | 0.02 | asserted 1.00 |
| s1 | While the evaluation component is vital, the accuracy of move value estimates is also fundamentally linked to how well the sampling distribution corresponds the true distribution. | background 0.98 | 0.08 | 0.03 | asserted 1.00 |
| s2 | Despite this, recent work in trick-taking card game AI has mainly focused on improving evaluation algorithms with limited work on improving sampling. | gap 0.99 | 0.05 | 0.03 | asserted 1.00 |
| s3 | In this paper, we focus on the effect of sampling on the strength of a player and propose a novel method of sampling more realistic states given move history. | contribution 1.00 | 0.80 | 0.14 | asserted 0.84 |
| s4 | In particular, we use predictions about locations of individual cards made by a deep neural network --- trained on data from human gameplay - in order to sample likely worlds for evaluation. | method 0.99 | 0.89 | 0.06 | asserted 0.98 |
| s5 | This technique, used in conjunction with Perfect Information Monte Carlo (PIMC) search, provides a substantial increase in cardplay strength in the popular trick-taking card game of Skat. | result 1.00 | 0.92 | 0.86 | asserted 1.00 |
What was examined, and what was not
- arXiv listing abstract v1
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
The paper
Original abstract
In trick-taking card games, a two-step process of state sampling and evaluation is widely used to approximate move values. While the evaluation component is vital, the accuracy of move value estimates is also fundamentally linked to how well the sampling distribution corresponds the true distribution. Despite this, recent work in trick-taking card game AI has mainly focused on improving evaluation algorithms with limited work on improving sampling. In this paper, we focus on the effect of sampling on the strength of a player and propose a novel method of sampling more realistic states given move history. In particular, we use predictions about locations of individual cards made by a deep neural network --- trained on data from human gameplay - in order to sample likely worlds for evaluation. This technique, used in conjunction with Perfect Information Monte Carlo (PIMC) search, provides a substantial increase in cardplay strength in the popular trick-taking card game of Skat.
Human review
How the literature uses it
2 arXiv papers cite this in 3 sentences
Papers citing it · 1 sentence without a confident label
Compares with it 1 paper
-
It also offers a controlled counterpoint to the learning-based body of work on trick-taking card games [ref] : the gap between a deterministic policy and a learned policy such as the one reported by [ref] is not primarily a gap in briscola management but a gap in the handling of the draw sequence, which is exactly the ingredient that Information-Set MCTS [ref] is able to reason about via opponent-inference and that a deterministic rule-based policy cannot recover.
Beyond the briscola advantage: a Monte Carlo dominance test for deterministic strategies in two-player Briscola Game · Discussion · 0.95
Cites it as background 1 paper
-
There have been larger efforts to apply machine learning to predict bidding options and hand cards in Skat [ref] .
Knowledge-Based Paranoia Search in Trick-Taking · Related Work · 0.98
What it cites
Uses the method or tool of 1 paper
-
Hidden layers use ELU activations [ref] , and the network is trained by minimizing the average cross-entropy of each card output.
Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs) · Application to Skat · 0.99
Discussion
Know something about this paper? Leave a comment, or highlight any passage above to comment on it.