Towards understanding the importance of time-series features in automated algorithm performance prediction.
Gasper PetelinGjorgjina CenikjTome EftimovPublished in: Expert Syst. Appl. (2023)
Keyphrases
- dynamic programming
- learning algorithm
- optimization algorithm
- cost function
- prediction algorithm
- detection algorithm
- segmentation algorithm
- worst case
- objective function
- preprocessing
- matching process
- experimental evaluation
- np hard
- k means
- extracted features
- computational complexity
- feature vectors
- optimal solution
- lower bound
- recognition algorithm
- times faster
- prediction error
- neural network
- prediction model
- convergence rate
- classification method
- genetic algorithm
- decision trees
- feature extraction
- expectation maximization
- high accuracy
- similarity measure
- computational cost
- search space
- artificial neural networks