search

kdmalc / personalization-privacy-risk

Privacy analysis for ML and classical filtering personalization parameters
0 stars 0 forks source link

Why is init cost so low? Random init dec should be the worst performing #21

Closed kdmalc closed 1 year ago

kdmalc commented 1 year ago

Theory is that it could be because we are using sum instead of norm for speed reasons (real time requirement) and PCA greatly reduces the sum (doesn't change the norm necessarily), thus the low init costs.

kdmalc commented 1 year ago

This got fixed, I forget why/how. I think changing from sum to norm did fix this? It was around then