Wednesday, June 26, 2019

LDA(半)深究

看了下LDA
本来是知道个大概,今天深入的看了一部分,还没完全搞懂。
基本他是由几个conditional probability 合起来
但是要对这个conditional probability 求解不容易
于是需要用到Gibb Sampling

另LDA 用途很广,图像什么也用到。
这东西有意思。
明天继续看。

Paper - Pivoting to topic modeling related
Pru - 0
Talk -
1. Think before speak, give good connection
2. Resort to emotion/entertainment/story telling

No comments:

Post a Comment