Pairing character classes in a deathmatch shooter game via a deep-learning surrogate model

Abstract

This paper introduces a surrogate model of gameplay that learns the mapping between different game facets, and applies it to a generative system which designs new content in one of these facets. Focusing on the shooter game genre, the paper explores how deep learning can help build a model which combines the game level structure and the game's character class parameters as input and the gameplay outcomes as output. The model is trained on a large corpus of game data from simulations with artificial agents in random sets of levels and class parameters. The model is then used to generate classes for specific levels and for a desired game outcome, such as balanced matches of short duration. Findings in this paper show that the system can be expressive and can generate classes for both computer generated and human authored levels. CCS CONCEPTS • Applied computing → Computer games; • Theory of computation → Evolutionary algorithms; • Computing methodologies → Neural networks;

10 Years of the PCG workshop: Past and Future Trends

“…The generation of mechanics is equally popular (8% of papers), sometimes combined with level generation [25] or graphics generation [64]. Mechanics refer to actions players can take [25,110], but in this case a broader view of mechanics is used to include gameplay parameters of weapons and character classes [45,64,69]. Generated story elements also feature in 5 papers, although the classification is broad since papers range from generating plots for mysteries [50], to NPC biographies [33], to satellite sentences [71].…”

“…In the last decade, 14 papers published at the PCG workshop use artificial evolution to generate levels (8 papers) and/or mechanics (4 papers), or graphics [19,37,64]. Machine learning has also been applied in 7 PCG workshop papers, in the form of non-negative matrix factorization [109], random forests [5], self-organizing maps [72], deep learning [44,45], and wave function collapse [46,47]. Notably, 4 of these 7 papers are published from 2017 onwards.…”

“…As noted in Section 5, machine learning (ML) is the most upwards trending algorithmic method with 4 of 7 ML papers published from 2017 onwards. More importantly, all four of these latest ML papers combine machine learning with artificial evolution [45,49] or declarative programming [46,47]. The heightened interest of PCG workshop publications in machine learning are likely attributed to recent successes in deep learning more broadly, the availability of easy-touse codebases for machine learning, and a game-specific academic interest in PCG via Machine Learning [90].…”

A literature review of Empirical Evidence on Procedural Content Generation in Game-Related Implementation

“…Platformer game utilizes a platform as a playing field for the player to do its actions, due to its nature, PCG are able to alter the playing field by using multiple methods [52]. First Person Shooter [92], [93] Another prevalent game model that uses maps are first person shooter (FPS) games, where a player controls an in-game avatar that has a first-person view as if the player is seeing the object in real life. Similar to platformer games, maps can also be generated using PCG [92].…”

“…First Person Shooter [92], [93] Another prevalent game model that uses maps are first person shooter (FPS) games, where a player controls an in-game avatar that has a first-person view as if the player is seeing the object in real life. Similar to platformer games, maps can also be generated using PCG [92]. Not limited to maps, FPS require players to face another player as an opponent, be it another human player, or an A.I., which in fact, can be generated procedurally by using PCG [93].…”