Selective Sliced Wasserstein Distance: An Efficient Approach for High-Dimensional Distribution Comparison
Imagine comparing two galaxies, not just by star count, but by their overall shape and distribution. This is analogous to comparing probability distributions, especially in high-dimensional spaces where each data point has numerous coordinates. Traditional methods, like the Wasserstein distance, can be computationally expensive, like measuring every interstellar distance. The Selective Sliced Wasserstein Distance (SSWD) … Read more