Hey everyone,
I am working on tuning hyperparameters for a machine learning model and wondering if convex optimization techniques can help streamline the process. Specifically…, I have a loss function that seems convex in some parameters but non-convex overall.
Are there any convex relaxation methods that can approximate a non-convex loss function effectively: ??
Would using projected gradient descent or other convex optimization techniques be beneficial in finding optimal hyperparameters: ??
Any recommended convex solvers or libraries that integrate well with ML frameworks like TensorFlow or PyTorch: ??
I would appreciate any insights or references to research papers that discuss convex optimization applications in ML hyperparameter tuning.
Thanks !!