***********************************************************************************

******************************************
WHY TSNEA CANT BE USED FOR CLASSIFICATION TASK?

In MNIST dataset, when a new image of 28*28 pixel is given, how can it be
projected on to 2D properly?
It is stochastic in nature so different coordinates different time. Even if we
assigned a random state, how is it possible to take a new 28*28 pixel image to 2D
with a single data point?
There will be no Neighborhood Embedding right? So we can't use t-SNE for
classification of new data point AT ALL. Am I correct?
***********************************************************************************
**************************************8
***********************************************************************************
**********************************

why training error is not zero?

****************************************************************************
why training error increases with increase in "k"

***********************************************************************************
***********************
In k-means or kNN, we use euclidean distance to calculate the distance between
nearest neighbours.
Why not manhattan distance ?
(https://www.analyticsvidhya.com/blog/2017/09/30-questions-test-k-nearest-
neighbors-algorithm/)
***********************************************************************************
*****************
How is kNN different from k-means clustering?
(https://stats.stackexchange.com/questions/56500/what-are-the-main-differences-
between-k-means-and-k-nearest-neighbours)

***********************************************************************************
***********************