Correct me if I'm wrong, but as per your (by the way excellent) explanations about the drawbacks of t-SNE, it would be incorrect to use the output of t-SNE as an input to a clustering algorithm, right? As you explained that t-SNE expands denser clusters & contracts sparse ones, and secondly, the distance between the clusters don't mean anything, so we can not aim to cluster the t-SNE output & check for visible clusters. Am I thinking correct?
Thank you so much sir for sharing this brilliant video for explaining logic of t-SNE. It is veryyyyy helpful.
Love love love thanks a lot... wonderful explanations👍...watched your full playlist on DR.
Back here again. Litsening daily.
Good one. but see at 1.5x or 2x
Very good explanation!
Some claim that perplexity is not a very reliable hyper parameter.
excellent
Great, Thank you
what software are you using for recording this?
Very nice
Correct me if I'm wrong, but as per your (by the way excellent) explanations about the drawbacks of t-SNE, it would be incorrect to use the output of t-SNE as an input to a clustering algorithm, right?
As you explained that t-SNE expands denser clusters & contracts sparse ones, and secondly, the distance between the clusters don't mean anything, so we can not aim to cluster the t-SNE output & check for visible clusters. Am I thinking correct?
really wanted to know the answer to this
I think you can not predict the class of a new data point because of the stochastic aspect. But the clusters are meaningful. Others can comment too.
you are so funny ahah
great tutorial!!
saves my ass for exams!