Hi!
I’ve been using openfhe to implement logistic regression training. I block the plaintext and pack each piece into a ciphertext. But in the process of training, I encountered a problem. It says: "lbcrypto::math_error: The decryption failed because the approximation error is too high. Check the parameters. "What parameters are causing the problem and what changes should I make?Here are the parameters I set.
Thanks for your advice. I have some problems running the lr_nag program in [ GitHub - openfheorg/openfhe-logreg-training-examples: OpenFHE-Based Examples of Logistic Regression Training using Nesterov Accelerated Gradient Descent](https:// GitHub - openfheorg/openfhe-logreg-training-examples: OpenFHE-Based Examples of Logistic Regression Training using Nesterov Accelerated Gradient Descent). When the code runs to
How much RAM does you machine have? CKKS bootstrapping typically works with large keys. You can try to use a smaller ring dimension (twice smaller).
Generally speaking, Approximation error is too high means that the decryption result is incorrect. This may happen for various reasons, e.g., incorrect use of keys or anything else.