Embedding doc #424
Embedding doc #424
Conversation
… documentations. Update loading animation
visualdl/python/storage.py
Outdated
|
|
||
| def embedding(self): | ||
| """ | ||
| Create an embedding writer that used to write |
visualdl/python/storage.py
Outdated
| Create an embedding writer that used to write | ||
| embedding data. | ||
| :return: A embedding writer to record embedding data |
visualdl/server/lib.py
Outdated
| return _handler | ||
|
|
||
|
|
||
| # A simple PCA implementaiton to do the dimension reduction. |
There was a problem hiding this comment.
Normally we comment methods like this:
def kos_root():
"""Return the pathname of the KOS root directory."""
global _kos_root
if _kos_root: return _kos_root
...
| cov = np.cov(x, rowvar=False) | ||
|
|
||
| # Get eigenvectors and eigenvalues from the covariance matrix | ||
| eigvals, eigvecs = np.linalg.eig(cov) |
There was a problem hiding this comment.
What a math guy, Jeff! But do we need to do it ourselves? Moreover, SVD is more stable than eigenvector in terms of computing principle components.
Check here
There was a problem hiding this comment.
The reason behind it so we don't have to import another pip package to do the calculation.
And actually the TSNE file already has a clean PCA implementation. and I could just use that.
| @@ -0,0 +1,184 @@ | |||
| # | |||
There was a problem hiding this comment.
We should give the original author a citation 😀
https://github.com/bhauman/neurpy/blob/master/tsne.py
There was a problem hiding this comment.
I did give the author credit, I left my comments under the author's comments.
3 participants
Reveal the high dimensional tab
Add function documentation
Improve loading animation algorithm
Use the in-house PCA function
Use the free T-SNE function
Resolve #407