Code and BeyondCode😋
Posts Tools Tags Categories BeyondCode😋
Code and BeyondCode😋

Contents

L3-Node Embeddings

 included in Stanford-CS224W
   34 words   One minute 

Node Embeddings

https://web.stanford.edu/class/cs224w/slides/02-nodeemb.pdf

/l3-node-embeddings/image.png

encoder and decoder

/l3-node-embeddings/image-1.png

encoder: simple example

/l3-node-embeddings/image-2.png ？？注意这里矩阵是one column per node， 这里似乎解释通了为什么glidar里面node在encode的过程中数量不变，换句话说就是 not scalable

呼之欲出啊啊啊啊啊 😱

以下内容非常具有启发性

/l3-node-embeddings/image-3.png

/l3-node-embeddings/image-4.png

/l3-node-embeddings/image-5.png

Random walks

/l3-node-embeddings/image-6.png

/l3-node-embeddings/image-7.png 怎么理解高效率？

对特征学习的考量 /l3-node-embeddings/image-8.png 提出损失函数 /l3-node-embeddings/image-9.png

/l3-node-embeddings/image-10.png

用了一个近似来化简 （不约而同走到了noise-denoise） /l3-node-embeddings/image-11.png k在5~20之间！又是glidar的论文！

summary

/l3-node-embeddings/image-12.png

node2vec

/l3-node-embeddings/image-13.png /l3-node-embeddings/image-14.png

/l3-node-embeddings/image-15.png

embedding the entire graph

SKIP

Updated on 2025-01-27
Back | Home