Neural Cellular Automata (NCA) Evolution Dataset
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/caitlingrasso/empowered-nca
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了25次神经网络细胞自动机（Neural Cellular Automata）的进化运行结果，这些运行旨在优化形态发生和赋能目标。数据集还包括了在使用不同目标函数训练NCAs时的性能信息，这些目标函数包括双损失和三损失处理，及其对形状匹配和赋能的影响。该数据集的规模涉及25次运行，种群大小为400个NCAs，任务集中在形态发生和赋能优化。
This dataset encompasses the results of 25 evolutionary runs of Neural Cellular Automata (NCA), which were conducted to optimize morphogenesis and empowerment-related goals. It also includes performance metrics obtained when training NCAs using diverse objective functions, namely dual-loss and triple-loss formulations, along with their respective impacts on shape matching and empowerment. Comprising 25 runs with a population size of 400 NCAs, this dataset centers its tasks on morphogenesis and empowerment optimization.
提供机构：
Vermont Advanced Computing Core
