Timeline for Can t-SNE be directly used as a clustering algorithm?
Current License: CC BY-SA 4.0
3 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Feb 7, 2020 at 6:51 | comment | added | machine-building | Hi Jan, I have doubts about the first two points raised: 1. What if the the clusters are only/more visible from a 2D embedding? 2. The same can be said to most unsupervised clustering algorithm (K-Means may result in noise clusters or disregard clusters that do exist as well). I'm more convinced by these two points: 1. I do note that the stochastic nature of t-SNE may affect reproducibility. 2. Behaviours of t-SNE can seem 'magical' at times as shown in How to Use t-SNE Effectively, which affects the explainability and credibility of the results. | |
| Feb 3, 2020 at 2:36 | vote | accept | machine-building | ||
| Jan 31, 2020 at 9:05 | history | answered | Jan Kukacka | CC BY-SA 4.0 |