Evolving indirectly encoded convolutional neural networks to play tetris with low-level features

Schrum, Jacob

doi:10.1145/3205455.3205459

Cited by 9 publications

(3 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There, Tetris has been used as a research tool more often than any other video game [115]. Tetris has be subjected to just about every machine learning and artificial intelligence paradigm [21,139,122,60,22,36,146,147,54,143,145,81,112,111,75,53,102,129,113,128,114,68,144,124,49,23,103,98,125,142,131,24,5,88,127,82,123] including, of course, ant colonies [32]. However, success is far from guaranteed, as The Unsuitability of Supervised Backpropagation Networks for Tetris shows [84].…”

Section: Related Workmentioning

confidence: 99%

How Fast Can We Play Tetris Greedily With Rectangular Pieces?

Dallant¹,

Iacono²

2022

Preprint

View full text Add to dashboard Cite

Consider a variant of Tetris played on a board of width w and infinite height, where the pieces are axis-aligned rectangles of arbitrary integer dimensions, the pieces can only be moved before letting them drop, and a row does not disappear once it is full. Suppose we want to follow a greedy strategy: let each rectangle fall where it will end up the lowest given the current state of the board. To do so, we want a data structure which can always suggest a greedy move. In other words, we want a data structure which maintains a set of O(n) rectangles, supports queries which return where to drop the rectangle, and updates which insert a rectangle dropped at a certain position and return the height of the highest point in the updated set of rectangles. We show via a reduction to the Multiphase problem [Pătraşcu, 2010] that on a board of width w = Θ(n), if the OMv conjecture [Henzinger et al., 2015] is true, then both operations cannot be supported in time O(n 1/2− ) simultaneously. The reduction also implies polynomial bounds from the 3-SUM conjecture and the APSP conjecture. On the other hand, we show that there is a data structure supporting both operations in O(n 1/2 log 3/2 n) time on boards of width n O(1) , matching the lower bound up to a n o(1) factor.

show abstract

Section: Related Workmentioning

confidence: 99%

How Fast Can We Play Tetris Greedily With Rectangular Pieces?

Dallant¹,

Iacono²

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Since there is no clear relationship between accuracy and diversity of an ensemble [12], the use of NSGA-II makes these objectives to be considered separately, thus maintaining the richness of both criteria in the evolution process. The final result is still the prediction accuracy as it is the main objective of interests [22,26]. Therefore, we simply choose the chromosome with the best accuracy from the last generation of NSGA-II as the final selected individual.…”

Section: Algorithmsmentioning

confidence: 99%

Multi-layer heterogeneous ensemble with classifier and feature selection

Nguyen

Pham

Dang

et al. 2020

Proceedings of the 2020 Genetic and Evolutionary Computation Conference

View full text Add to dashboard Cite

Deep Neural Networks have achieved many successes when applying to visual, text, and speech information in various domains. The crucial reasons behind these successes are the multi-layer architecture and the in-model feature transformation of deep learning models. These design principles have inspired other sub-fields of machine learning including ensemble learning. In recent years, there are some deep homogenous ensemble models introduced with a large number of classifiers in each layer. These models, thus, require a costly computational classification. Moreover, the existing deep ensemble models use all classifiers including unnecessary ones which can reduce the predictive accuracy of the ensemble. In this study, we propose a multi-layer ensemble learning framework called MUlti-Layer heterogeneous Ensemble System (MULES) to solve the classification problem. The proposed system works with a small number of heterogeneous classifiers to obtain ensemble diversity, therefore being efficiency in resource usage. We also propose an Evolutionary Algorithm-based selection method to select the subset of suitable classifiers and features at each layer to enhance the predictive performance of MULES. The selection method uses NSGA-II algorithm to optimize two objectives concerning classification accuracy and ensemble diversity. Experiments on 33 datasets confirm that MULES is better than a number of well-known benchmark algorithms. CCS CONCEPTS • Computing methodologies → Ensemble methods; • Mathematics of computing → Evolutionary Algorithms;

show abstract

“…The original Tetris TM has seven different types of zoids (Figure 2) and takes place on a board of 20x10 blocks. Each zoid consists of 4-connected blocks, that is, each block of the zoid is connected to at least one other block in one of the four 2 Tetris TM has been used for several objectives, like training of spatial skills 30 (Milani et al, 2019), analysis of cognitive abilities like cognitive workload (Trithart, 3 2000), as an investigation tool to investigate mental processes linked to pragmatic actions and epistemic action (Kirsh & Maglio, 1994), or as a work-space in which to train and test neural models or other AI algorithms able to compete or reproduce human performance (Schrum, 2018;Lora Ariza et al, 2017).…”

Section: Introductionmentioning

confidence: 99%

The role of mental rotation in TetrisTM gameplay: An ACT-R computational cognitive model

Gentile

Lieto

2022

Cognitive Systems Research

View full text Add to dashboard Cite

The mental rotation ability is an essential spatial reasoning skill in human cognition and has proven to be an essential predictor of mathematical and STEM skills, critical and computational thinking. Despite its importance, little is known about when and how mental rotation processes are activated in games explicitly targeting spatial reasoning tasks. In particular, the relationship between spatial abilities and Tetris TM has been analysed several times in the literature. However, these analyses have shown contrasting results between the effectiveness of Tetris-based training activities to improve mental rotation skills. In this work, we studied whether, and under what conditions, such ability is used in the Tetris TM game by explicitly modelling mental rotation via an ACT-R based cognitive model controlling a virtual agent. The obtained results show meaningful insights into the activation of mental rotation during game dynamics. The study suggests the necessity to adapt game dynamics in order to force the activation of this process and, therefore, can be of inspiration to design learning activities based on Tetris TM or re-design the game itself to improve its educational effectiveness.

show abstract

Evolving indirectly encoded convolutional neural networks to play tetris with low-level features

Cited by 9 publications

References 32 publications

How Fast Can We Play Tetris Greedily With Rectangular Pieces?

How Fast Can We Play Tetris Greedily With Rectangular Pieces?

Multi-layer heterogeneous ensemble with classifier and feature selection

The role of mental rotation in TetrisTM gameplay: An ACT-R computational cognitive model

Contact Info

Product

Resources

About