Open Thewindrisers opened 2 years ago
老哥你好,关于您分享的Federated Learning with Non-IID Data这篇文章,有个问题想请教一下,就是在EMD那块,EMD不是一个衡量分布之间差异的指标吗,作者为何行文中都是用E=1E=5这样描述,我不太明白他是怎么计算EMD的,如果是用概率分布计算,我用scipy.stats 的 wasserstein_distance计算EMD都是在0.1左右,想听听你的理解,谢谢。
老哥,你EMD的问题解决了吗?我目前也存在这问题。
老哥你好,关于您分享的Federated Learning with Non-IID Data这篇文章,有个问题想请教一下,就是在EMD那块,EMD不是一个衡量分布之间差异的指标吗,作者为何行文中都是用E=1E=5这样描述,我不太明白他是怎么计算EMD的,如果是用概率分布计算,我用scipy.stats 的 wasserstein_distance计算EMD都是在0.1左右,想听听你的理解,谢谢。