Using Data Augmentation and Machine Learning for Identifying the Correct Level of Difficulty in Games

• André Gomes da Silva IFRJ
• Fernando Santos PUC-Rio
• Augusto Baffa PUC-Rio
• Jose Ricardo da Silva Junior IFRJ

DOI: https://doi.org/10.5753/sbgames.2026.25306

Resumo

Introduction: Dynamic Difficulty Adjustment (DDA) aims to balance player experience by adapting game challenge to individual abilities. Traditional approaches rely either on static difficulty selection or data-intensive adaptive models, both of which present limitations regarding player self-assessment and data availability. Objective: This work proposes a machine learning approach to identify the most appropriate difficulty level for games based on player performance metrics and subjective perceptions, while addressing the challenge of limited training data. Methodology or Steps: We model the problem as a three-class classification task (Increase, Keep, Decrease difficulty) using Logistic Regression. We adopt a CTGAN-based data augmentation approach to generate synthetic player data, thereby addressing data scarcity. The approach is evaluated using gameplay data from Doom II sessions, incorporating both objective performance metrics and players’ perceived difficulty. Results: Experimental results show that models trained with synthetic data improve predictive performance compared to models trained only on real data, achieving higher F1-scores and better generalization. The findings suggest that synthetic data can enhance learning in low-data scenarios, although a domain gap between synthetic and real data persists, particularly when distinguishing between similar behavioral classes.

Palavras-chave: Dynamic Difficulty Adjustment, Player Modeling, Support Vector Classification, Data Augmentation, Synthetic Data, Game Analytics

Referências

Afergan, D., Peck, E. M., Solovey, E. T., Jenkins, A., Hincks, S. W., Brown, E. T., Chang, R., e Jacob, R. J. (2014). Dynamic difficulty using brain metrics of workload. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, CHI ’14, page 3797–3806, New York, NY, USA. Association for Computing Machinery.

Alexander, J. T., Sear, J., e Oikonomou, A. (2013). An investigation of the effects of game difficulty on player enjoyment. Entertainment Computing, 4(1):53–62.

Bicalho, L. F., Baffa, A., e Feijó, B. (2024). A dynamic difficulty adjustment algorithm with generic player behavior classification unity plugin in single player games. In Proceedings of the 22nd Brazilian Symposium on Games and Digital Entertainment, SBGames ’23, page 76–85, New York, NY, USA. Association for Computing Machinery.

Csikszentmihalyi, M. e Csikszentmihalyi, I. S. (1992). Optimal Experience: Psychological Studies of Flow in Consciousness. Cambridge University Press.

Davis, J. P., Steury, K., e Pagulayan, R. (2005). A survey method for assessing perceptions of a game: The consumer playtest in game design. Game Studies 5.

Figueira, F. M., Nascimento, L., da Silva Junior, J., Kohwalter, T., Murta, L., e Clua, E. (2018). Bing: A framework for dynamic game balancing using provenance. In 2018 17th Brazilian Symposium on Computer Games and Digital Entertainment (SBGames), pages 57–5709.

Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., e Bengio, Y. (2020). Generative adversarial networks. Commun. ACM, 63(11):139–144.

Hawkins, G., Nesbitt, K., e Brown, S. (2012). Dynamic difficulty balancing for cautious players and risk takers. Int. J. Comput. Games Technol., 2012:3:3–3:3.

Hunicke, R. (2005). The case for dynamic difficulty adjustment in games. In Proceedings of the 2005 ACM SIGCHI International Conference on Advances in Computer Entertainment Technology, ACE ’05, pages 429–433, New York, NY, USA. ACM.

id Software (1994). Doom II: Hell on earth. MS-DOS. Jogo eletrônico.

Kristensen, J. T., Guckelsberger, C., Burelli, P., e Hämäläinen, P. (2022). Personalized game difficulty prediction using factorization machines. In Proceedings of the 35th Annual ACM Symposium on User Interface Software and Technology, UIST ’22, New York, NY, USA. Association for Computing Machinery.

Li, X., He, S., Dong, Y., Liu, Q., Liu, X., Fu, Y., Shi, Z., e Huang, W. (2010). To create dda by the approach of ann from uct-created data. In 2010 International Conference on Computer Application and System Modeling (ICCASM 2010), volume 8, pages V8– 475–V8–478.

Liu, C., Agrawal, P., Sarkar, N., e Chen, S. (2009). Dynamic difficulty adjustment in computer games through real-time anxiety-based affective feedback. International Journal of Human–Computer Interaction, 25(6):506–529.

Lopes, P., Fachada, N., e Fonseca, M. (2025). Closing the loop: A systematic review of experience-driven game adaptation.

Melhart, D., Azadvar, A., Canossa, A., Liapis, A., e Yannakakis, G. N. (2019). Your gameplay says it all: Modelling motivation in tom clancy’s the division. In 2019 IEEE Conference on Games (CoG), pages 1–8.

Melhart, D., Liapis, A., e Yannakakis, G. N. (2022). The arousal video game annotation (again) dataset. IEEE Transactions on Affective Computing, 13(4):2171–2184.

Missura, O. e Gärtner, T. (2011). Predicting dynamic difficulty. In Shawe-Taylor, J., Zemel, R. S., Bartlett, P. L., Pereira, F., e Weinberger, K. Q., editors, Advances in Neural Information Processing Systems 24, pages 2007–2015. Curran Associates, Inc.

Mourato, F., Birra, F., e dos Santos, M. P. (2014). Difficulty in action based challenges: Success prediction, players’ strategies and profiling. In Proceedings of the 11th Conference on Advances in Computer Entertainment Technology, ACE ’14, pages 9:1– 9:10, New York, NY, USA. ACM.

Müller, A. C. e Guido, S. (2016). Introduction to Machine Learning with Python: A Guide for Data Scientists. O’Reilly Media, Inc.

Olesen, J. K., Yannakakis, G. N., e Hallam, J. (2008). Real-time challenge balance in an rts game using rtneat. In 2008 IEEE Symposium On Computational Intelligence and Games, pages 87–94.

Patki, N., Wedge, R., e Veeramachaneni, K. (2016). The synthetic data vault. In IEEE International Conference on Data Science and Advanced Analytics (DSAA), pages 399–410.

Pelling, C. e Gardner, H. (2019). Two human-like imitation-learning bots with probabilistic behaviors. In 2019 IEEE Conference on Games (CoG), pages 1–7.

Pérez, L. J. F., Calla, L. A. R., Valente, L., Montenegro, A. A., e Clua, E. W. G. (2015). Dynamic game difficulty balancing in real time using evolutionary fuzzy cognitive maps. In 2015 14th Brazilian Symposium on Computer Games and Digital Entertainment, pages 24–32.

Rashed, A., Shirmohammadi, S., e Hefeeda, M. (2025). Descriptor: Multimodal dataset for player engagement analysis in video games (multipeng). IEEE Data Descriptions, 2:17–25.

Rietveld, A., Bakkes, S., e Roijers, D. (2014). Circuit-adaptive challenge balancing in racing games. In 2014 IEEE Games Media Entertainment, pages 1–8.

Rosa, M. P. C., dos Santos, E. A., de Moraes, I. L. R., e Silva, T. B. P., Sarmet, M. M., Castanho, C. D., e Jacobi, R. P. (2021). Dynamic difficulty adjustment using performance and affective data in a platform game. In Stephanidis, C., Soares, M. M., Rosenzweig, E., Marcus, A., Yamamoto, S., Mori, H., Rau, P.-L. P., Meiselwitz, G., Fang, X., e Moallem, A., editors, HCI International 2021 - Late Breaking Papers: Design and User Experience, pages 367–386, Cham. Springer International Publishing.

Schell, J. (2008). The Art of Game Design: A Book of Lenses. Morgan Kaufmann Publishers.

Sifa, R., Runge, J., Bauckhage, C., e Klapper, D. (2018). Customer lifetime value prediction in non-contractual freemium settings: Chasing high-value users using deep neural networks and smote.

Sifa, R. e Yang, E. (2023). A comparative analysis of data augmentation approaches for improved minority behavior detection in digital games. In 2023 IEEE International Conference on Big Data (BigData), pages 6275–6278.

Tran, T. D., Heinzig, M., Langner, H., e Ritter, M. (2025). Towards adaptive difficulty and personalized player experience: A data-driven approach combining game analytics, sensor data, and personality traits. In Proceedings of the Mensch Und Computer 2025, MuC ’25, page 29–46, New York, NY, USA. Association for Computing Machinery.