Sequential Segment-based Level Generation and Blending using Variational Autoencoders

Sarkar, Anurag; Cooper, Seth

doi:10.1145/3402942.3409604

Cited by 31 publications

(39 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Some prior work has modeled Mega Man with machine learning without generating new Mega Man content [7,12]. Sarkar et al have modeled Mega Man levels along with levels from a large number of other games with Variational Autoencoders (VAEs) for the purpose of recombining this content to create entirely new types of content [14][15][16][17]33]. We instead focus on the problem of generating levels that resemble those from the original Mega Man.…”

Section: Related Workmentioning

confidence: 99%

“…To the best of our knowledge, the only prior PCGML work that has incorporated ensemble learning has been work that employs a Random Forest (RF). However, all prior instances that have employed RFs have done so for a secondary classification task, and not for the primary generation task [5,14,17]. As such, our work stands out as the first PCGML approach to employ an ensemble of simple models for the primary generation task.…”

Section: Related Workmentioning

confidence: 99%

“…The current, common solutions to dealing with high variance in PCGML still employ a single model. When confronted with high variance many researchers focus on modeling a sub-problem, for example training a model to generate subsections of levels [1,14] or by employing a more abstract representation [30]. Alternatively, some researchers have used fitness functions to attempt to guide the machine learned model's generation when the available training data does not allow for sufficient generalization alone [31,32].…”

mentioning

confidence: 99%

See 2 more Smart Citations

Ensemble Learning For Mega Man Level Generation

Li,

Chen,

Xue

et al. 2021

Preprint

View full text Add to dashboard Cite

Procedural content generation via machine learning (PCGML) is the process of procedurally generating game content using models trained on existing game content. PCGML methods can struggle to capture the true variance present in underlying data with a single model. In this paper, we investigated the use of ensembles of Markov chains for procedurally generating Mega Man levels. We conduct an initial investigation of our approach and evaluate it on measures of playability and stylistic similarity in comparison to a non-ensemble, existing Markov chain approach. CCS CONCEPTS• Computing methodologies → Ensemble methods.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation

Ensemble Learning For Mega Man Level Generation

Li,

Chen,

Xue

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…We employ a Random Forest (RF), as a low-data classifier to approximate a "human-like" fitness function. Random Forests have appeared in prior PCGML work [6,16,18]. The major difference is that in this prior work, the RF is trained once on existing data, while we iteratively train our RF on generated data to approximate our desired fitness function.…”

Section: Procedural Content Generation Via Machine Learningmentioning

confidence: 99%

Adversarial Random Forest Classifier for Automated Game Design

Maurer

Guzdial

2021

The 16th International Conference on the Foundations of Digital Games (FDG) 2021

View full text Add to dashboard Cite

Autonomous game design, generating games algorithmically, has been a longtime goal within the technical games research field. However, existing autonomous game design systems have relied in large part on human-authoring for game design knowledge, such as fitness functions in search-based methods. In this paper, we describe an experiment to attempt to learn a human-like fitness function for autonomous game design in an adversarial manner. While our experimental work did not meet our expectations, we present an analysis of our system and results that we hope will be informative to future autonomous game design research.

show abstract

“…While generating whole levels by stitching together successively generated segments [13] is acceptable for a game like Mario with uniform progression, this does not work for games that progress in multiple directions and orientations. This was addressed in [14] via an approach combining a VAE-based sequential model and a classifier to generate and then place a segment relative to the previous one, thus generating whole levels by an iterative loop of decoding and encoding successive segments. However this afforded control only via definition of the initial segment with the orientation of successive segments and properties of the blend not being controllable.…”

Section: Introductionmentioning

confidence: 99%

Dungeon and Platformer Level Blending and Generation using Conditional VAEs

Sarkar¹,

Cooper²

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Variational autoencoders (VAEs) have been used in prior works for generating and blending levels from different games. To add controllability to these models, conditional VAEs (CVAEs) were recently shown capable of generating output that can be modified using labels specifying desired content, albeit working with segments of levels and platformers exclusively. We expand these works by using CVAEs for generating whole platformer and dungeon levels, and blending levels across these genres. We show that CVAEs can reliably control door placement in dungeons and progression direction in platformer levels. Thus, by using appropriate labels, our approach can generate whole dungeons and platformer levels of interconnected rooms and segments respectively as well as levels that blend dungeons and platformers. We demonstrate our approach using The Legend of Zelda, Metroid, Mega Man and Lode Runner.

show abstract

Sequential Segment-based Level Generation and Blending using Variational Autoencoders

Cited by 31 publications

References 10 publications

Ensemble Learning For Mega Man Level Generation

Ensemble Learning For Mega Man Level Generation

Adversarial Random Forest Classifier for Automated Game Design

Dungeon and Platformer Level Blending and Generation using Conditional VAEs

Contact Info

Product

Resources

About