Hittauppsatser.se

Uppsats

Using neural networks to predict winning strategies in spatio-temporal rock-paper-scissors games

Kandidat-uppsats

Jansson, Albert, Ullenhag, Lukas

Publicerad: 2025

Språk: Engelska

Nyckelord

klicka för att söka
⌕Artificial Intelligence⌕machine learning⌕AI⌕Artificiell intelligens⌕maskininlärning⌕Deep learning⌕Djupinlärning⌕Supervised Learning⌕Neural Networks⌕Neurala Nätverk⌕Classification⌕CNN⌕Convolutional Neural Network⌕klassificering⌕Evolutionary Game Theory⌕EGT⌕Rock⌕paper⌕Scissors⌕Evolutionär spelteori⌕sten⌕Sax⌕Påse

Sammanfattning

Predicting population dynamics over time is crucial for numerous scientific fields, including cancer research. Evolutionary Game Theory is commonly used to predict population dynamics over time. While analytical solutions exist for homogenous systems, simulations are often the only method for predicting population dynamics in spatially inhomogeneous systems. This project aims to explore an alternative machine-learning approach to predicting population dynamics in spatially inhomogeneous systems by using Convolutional Neural Networks to predict the winning strategy in simulated, spatial, Rock-Paper-Scissors games. The problem is constructed as a classification task, where simulation outcomes are sorted into five categories. A CNN model capable of predicting the correct outcome with 65 % accuracy is developed, outperforming traditional, analytical methods. The method shows promise and could be worth expanding on.

Information

Författare
Jansson, Albert, Ullenhag, Lukas
Publiceringsdatum
2025
Uppsatstyp
Kandidat-uppsats
Språk
Engelska
Nyckelord
⌕Artificial Intelligence⌕machine learning⌕AI⌕Artificiell intelligens⌕maskininlärning⌕Deep learning⌕Djupinlärning⌕Supervised Learning⌕Neural Networks⌕Neurala Nätverk⌕Classification⌕CNN⌕Convolutional Neural Network⌕klassificering⌕Evolutionary Game Theory⌕EGT⌕Rock⌕paper⌕Scissors⌕Evolutionär spelteori⌕sten⌕Sax⌕Påse

Utforska vidare

Liknande uppsatser

Uppsatser med liknande ämnen och nyckelord.