Excellent video. When applying tSNE to MNIST, are calculations of neighborhood get affect by the curse of dimensionality ? Considering that MNIST is a 28*28= 784 dimensional vector
@Applied AI Course , If we take a point from high dimensional and place it in low dimensional ,. How're we choosing our replaced points coz otherwise we'll have same number of points in low dimensional as well ,. If we replace all high dimensional points in low dimension
Our objective here is not to reduce the number of points, but to reduce the dimensionality of each point so that we can visualise the lower dimensional data.
if you handle your data as independent distributions, you can determine with confidence intervals those distributions and the total probability distributions, which histogram estimates in very high dimensions
