Do you mean principal component analysis? My naive understanding after reading the original paper is that the algorithm is training a transformation to project

Do you mean principal component analysis? My naive understanding after reading the original paper is that the algorithm is training a transformation to project the high dimensional data into low dimensions by best preserving both the global and local proximity. So that the samples similar at high dim space should also be close in the low dim one. It has some assumption of the distribution of the data in low dim so it won't be a random guess. It's using the t-distribution at low dim hence the name t-SNE. Correct me if any mistakes.