The impact of hyper-parameter tuning for landscape-aware performance regression and algorithm selection.
Anja JankovicGorjan PopovskiTome EftimovCarola DoerrPublished in: GECCO (2021)
Keyphrases
- parameter tuning
- learning algorithm
- optimal solution
- k means
- times faster
- preprocessing
- selection algorithm
- search space
- cost function
- np hard
- dynamic programming
- computational cost
- convergence rate
- parameter settings
- ink bleed
- segmentation algorithm
- expectation maximization
- high accuracy
- experimental evaluation
- image segmentation
- lower bound
- optimization algorithm
- matching algorithm
- computational complexity
- search algorithm
- objective function
- decision trees
- feature selection