SVC is slow on a medium-sized dataset and RAM is available. Which cache setting might help without changing the…
Classic MLLevel: IntermediateKernel cache performance
SVC is slow on a medium-sized dataset and RAM is available. Which cache setting might help without changing the objective?
Hints
- Kernel evaluations can be reused.
- A cache changes computation, not the optimization problem.
Answer
Increase cache_size to store more kernel computations, within the memory budget. It may reduce repeated computation, but cannot remove the fundamental scaling limits of kernel SVMs.
Source and licence: scikit-learn 1.7.2 — svm · BSD-3-Clause · answer: Quantoros editorial