Can Cebra be used for neuron clustering like tsne instead of neuron dynamic analysis？ #162
Replies: 1 comment 3 replies
|
Yes... we benchmark against it in our paper. |
3 replies
|
However, after dimensionality reduction, the data usually forms a circular distribution, which is not conducive to clustering. |
|
well, no it's highly dependent on the data and what you are using as auxiliary variables; ie see Figure 2: if you use only a discrete label (direction), you get discrete clusters: |
Answer selected by
MMathisLab
|
Then, if I use cebra for unsupervised clustering, can't I obtain discrete clusters? |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
well, no it's highly dependent on the data and what you are using as auxiliary variables; ie see Figure 2:
if you use only a discrete label (direction), you get discrete clusters: