Fractal-based complexity measures for predicting generalization in deep neural networks
收藏REDU2025-01-01 更新2026-05-11 收录
下载链接：
https://redu.unicamp.br/citation?persistentId=doi:10.25824/redu/EGL28M
下载链接
链接失效反馈官方服务：
资源简介：
Collection of materials for the thesis on generalization in deep neural networks. The tables folder contains CSV files with final tables (granular measures) and Kendall τ correlations by architecture settings. The code folder provides Jupyter notebooks for reading and saving CIFAR-10, used to train and save the models and evaluate the studied complexity measures.
提供机构：
. Instituto de Matemática Estatística e Ciência da Computação)
创建时间：
2025-01-01
