kNN.16 Locality sensitive hashing (LSH)

[@sourceIsDesire]

Hi Victor, Thanks for the great video! Just one question: how to you pick a "random" hyper-plane? I.e. which probability measures do you choose for which degrees of freedom? Cheers!

[@gabrieltakacs7193]

A simple way to choose a random hyper-plane in n-dimensions by choosing an n-dimensional vector consisting of n random values (independently sampled, say from a uniform or Gaussian distribution). If you treat this random vector as the normal to the hyper-plane, then you can decide which side of the hyper-plane you are on by looking at the sign of the inner product of the random normal and any point in the space.

[@LorenzoMella]

Oh no! We are entering the realm of Bertrand's Paradox with this! What's a "random" hyperplane? ;)

[@maryam_nn]

Very well explained! thank you!