at 47:28, why N is the lattice dimension? shouldn't it be the size of the vector？ Why in this CKKS case, N is 16? Looks like N is the size of vector which is 8, and in the pic, we need 3 times of rotation and get the sum is 36. isn't it?
at 32:24, the noise during the encryption, is it a side effect of encryption? or is it intended to be added forwardly? to be added for make encrypted data secure?
hi, at 25:10, you mean that after "cts[i].scale() = pow(2.0, 300)" , just this code generates an error in the result when decrypt? But it will be easier when the result scale is some powers of 2? at 33:57, so "cts[i].scale() = pow(2.0, 300)" is a brute force thing which causes some small error, right?
I'm working on one time pads with pictures using fonts, levels and cloaks. All good till the homebrew kicks in :-) job done.
at 47:28, why N is the lattice dimension? shouldn't it be the size of the vector？
Why in this CKKS case, N is 16?
Looks like N is the size of vector which is 8, and in the pic, we need 3 times of rotation and get the sum is 36. isn't it?
at 47:55, why N is 16, and N is the lattice dimension, and why the vector is half of that?
at 30:04, why 3 numbers, why not 4 numbers like {50,30,50,50}?
the Video only talks about why {50,30,50} but not covers why not {50,30,50,50}
at 32:24, the noise during the encryption, is it a side effect of encryption? or is it intended to be added forwardly? to be added for make encrypted data secure?
at 59:27, what is FFT? There's all kinds of errors coming from FFT.
hi, at 25:10, you mean that after "cts[i].scale() = pow(2.0, 300)" , just this code generates an error in the result when decrypt? But it will be easier when the result scale is some powers of 2?
at 33:57, so "cts[i].scale() = pow(2.0, 300)" is a brute force thing which causes some small error, right?
hi , what do you mean by message expansion?
from 43:41 rotation
hi at 19:34, how about the result p of the dot encode method? is this p also a 4096 vector or just one single plaintext value?
hi, at 17:03, how do you get the 218? how to calculate this maximum allowed coeff_modulus bit count?
Why is the video on CKKS not available?
at 1:05:21, galoiskeys caould be half if is created in a more intelligent way.
at 55:03, summation.
Thanks for defining "Microsoft Research" as "(MSR)". Too bad SEAL is not defined or described in the talk or video description
Worst explanation I have ever seen for someone's own work