TILDE-Q: A Transformation Invariant Loss Function for Time-Series Forecasting.
Hyunwook LeeChunggi LeeHongkyu LimSungahn KoPublished in: CoRR (2022)
Keyphrases
- loss function
- pairwise
- support vector
- geometric transformations
- logistic regression
- learning to rank
- risk minimization
- gradient boosting
- loss minimization
- regularization term
- boosting framework
- empirical risk
- reproducing kernel hilbert space
- stochastic gradient descent
- bayes rule
- affine transformation
- boosting algorithms
- feature space