Komparasi Algoritma WOA, MFO dan Genetic pada Optimasi Evolutionary Neural Network dalam Menyelesaikan Permainan 2048
DOI:
https://doi.org/10.17977/um068v1i92021p676-684Kata Kunci:
evolving neural network, genetic algorithm, woa, mfoAbstrak
Neural network optimization using evolutionary algorithms is an interesting research topic. But right now, there are not much research in this topic that focused on Game, especially 2048. The 2048 game is one of the interesting games to study considering that the level of difficulty of this game will increase when the value of the resulting number increases. In addition, this game is also not limited by time but can be played continuously until the game ends. Neural network and tree are 2 architectures that can be used to play 2048 but require a long training time if you want to play well. In this study, this problem was optimized by an evolutionary algorithm (3 algorithms used in this study: Genetic Algorithm, WOA, and MFO). With this optimization, the best weight will be obtained in either the NN or Tree architecture to produce good intelligence in playing 2048. After going through various trials, it is concluded that the combination with the NN architecture is better than the Tree architecture and the WOA and MFO algorithms have succeeded in optimizing the architecture with better than the genetic algorithm.
Optimasi neural network menggunakan algoritma evolutionary adalah topik penelitian yang menarik akan tetapi tidak banyak penelitian terkait hal ini yang berfokus pada game terutama game 2048. Game 2048 adalah salah satu game yang menarik untuk diteliti mengingat tingkat kesulitan permainan ini akan semakin meningkat disaat nilai angka yang dihasilkan semakin tinggi. Selain itu, permainan ini juga tidak dibatasi oleh waktu melainkan dapat dimainkan terus menerus hingga permainan berakhir. Neural network dan tree adalah 2 arsitektur yang dapat digunakan untuk memainkan 2048 akan tetapi membutuhkan waktu training yang lama jika ingin bermain dengan baik. Lama training tersebut yang pada penelitian ini dioptimasi oleh algoritma evolutionary (3 algoritma yang digunakan pada penelitian ini: Algoritma Genetic, WOA, dan MFO). Dengan adanya optimasi ini maka akan diperoleh bobot terbaik baik pada arsitektur NN ataupun Tree sehingga menghasilkan kecerdasan yang baik dalam memainkan 2048. Setelah melalui berbagai ujicoba maka disimpulkan bahwa kombinasi dengan arsitektur NN lebih baik dibandingkan dengan arsitektur Tree dan algoritma WOA dan MFO berhasil mengoptimasi arsitektur dengan lebih baik dibandingkan algoritma genetic.
Referensi
Boumaza, A. (2009). On the evolution of artificial Tetris players. The IEEE Symposium on Computational Boumaza, A. (2009, September). On the evolution of artificial Tetris players. In 2009 IEEE Symposium on Computational Intelligence and Games (pp. 387-393). IEEE.
Dickey, M. R. (2014). Puzzle Game 2048 Will Make You Forget Flappy Bird Ever Existed. Business Insider. Retrieved, 27
Ferri-Benedetti, F. (2014). The creator of 2048 tells us about the secret behind the game’s success. Retrieved from https://en.softonic.com/articles/interview-with-the-creator-of-2048.
Fogel, D. B. (1998). Evolutionary Computation. The Fossil Record. Selected Readings on the History of Evolutionary Computation. Classifier Systems.
Fogel, D. B., Fogel, L. J., & Porto, V. W. (1990). Evolving neural networks. Biological cybernetics, 63(6), 487-493.
Frank, K. D., Rich, C., & Longcore, T. (2006). Effects of artificial night lighting on moths. Ecological consequences of artificial night lighting, 13, 305-344.
Gaston, K. J., Bennie, J., Davies, T. W., & Hopkins, J. (2013). The ecological impacts of nighttime light pollution: a mechanistic appraisal. Biological reviews, 88(4), 912-927. https://doi.org/10.1111/brv.12036
Hintz, K. J., & Spofford, J. J. (1990, September). Evolving a neural network. In Proceedings. 5th IEEE International Symposium on Intelligent Control 1990 (pp. 479-484). IEEE. https://doi.org¬/10.¬110¬9/-IS¬IC.1990.128500
Mirjalili, S. (2015). Moth-flame optimization algorithm: A novel nature-inspired heuristic paradigm. Knowledge-based systems, 89, 228-249.https://doi.org/10.1016/j.knosys.2015.07.006
Mirjalili, S., & Lewis, A. (2016). The whale optimization algorithm. Advances in engineering software, 95, 51-67. /https://doi.org/10.1016/j.advengsoft.2016.01.008
Neller, T. W. (2015). Pedagogical possibilities for the 2048 puzzle game. Journal of Computing Sciences in Colleges, 30(3).
Olson, R. (2014). Artificial Intelligence has crushed all human records in 2048. Here’s how the AI pulled it off.
Turing, A. M. (2009). Computing machinery and intelligence (pp. 23-65). Springer Netherlands.
Watkins, W. A., & Schevill, W. E. (1979). Aerial observation of feeding behavior in four baleen whales: Eubalaena glacialis, Balaenoptera borealis, Megaptera novaeangliae, and Balaenoptera physalus. Journal of Mammalogy, 60(1), 155-163. https://doi.org/10.2307/1379766.
3.png)
1.png)
1.png)
