Matrix and vector dimensions in w2v

时间:2018-02-03 11:11:45

标签: word2vec word-embedding

I need an explanation of the dimensions of matrices and vectors in word2vec (CBOW algorithm). What they exactly are, why do we need them and what I should know about that?

Photo of CBOW model

1 个答案:

答案 0 :(得分:-1)

矩阵表示神经网络各层之间连接的权重。例如,W< NxV 矩阵具有N行,一个用于隐藏层的N个分量中的每一个,以及V列,一个用于输出层的每个V分量。它是一个矩阵,每个元素w' ij 表示隐藏层中第i个神经元与输出层中第j个神经元之间连接的权重。