Course talk:CPSC522/Concatenating Hyperspherical Distributions in Hyperspherical VAE
- [View source↑]
- [History↑]
Contents
Thread title | Replies | Last modified |
---|---|---|
Critique | 0 | 20:09, 21 April 2023 |
Critique 2 | 0 | 19:52, 21 April 2023 |
Critique | 0 | 01:48, 21 April 2023 |
I really liked reading the article. The motivation, experiments and discussion are written very well. A few comments,
- Add a line about batch effects and what regressing out batch effects mean.
- How do you decide the n in S_n to be 2, 5 and 40? and why not gradually increase it?
- Add wiki links to refer to the formula behind Adjusted Rand Index (ARI), Normalized Mutual Information (NMI) and Adjusted Mutual Information (AMI)
MEHARBHATIA (talk)
The article reads really well with all aspects being covered. I had a few minor points,
- Can you rephrase this line, “two high-dimensional cell datasets in a lower but still high-dimensional latent space” because its not clear if the latent space is lower or higher.
- Results do not include Adjusted Rand Index (ARI), Normalized Mutual Information (NMI) and Adjusted Mutual Information (AMI) for evaluation
- Zoom out/increase the size of the clustering figures, it is a bit difficult to see the legend.
NIKHILSHENOY (talk)
Overall, I found it an interesting experiment into improving hyperspherical VAEs in single-cell analyses.
My specific comments are here: https://drive.google.com/file/d/1E1aImedH-egCShIYgmQlRb7o91P2NFXv/view?usp=sharing