Expand description
Auto-generated module
🤖 Generated with SplitRS
Structs§
- AdaBoost
- AdaBoost: adaptive boosting with exponential loss.
- Bias
Variance Tradeoff - Bias-variance tradeoff decomposition: MSE = Bias² + Variance + Noise.
- Causal
Backdoor - Backdoor adjustment formula for causal inference.
- Cross
Validation - Cross-validation scheme.
- Double
Rademacher - Double Rademacher (two-sided) bound.
- ELBO
- Evidence Lower Bound (ELBO) for variational inference.
- Early
Stopping Reg - Early stopping regularization — implicit regularization by iteration count.
- Exponential
Weights Algorithm - Exponential Weights Algorithm (Hedge / EWA) for online learning.
- Feature
Map - Feature map: maps inputs to a (truncated) explicit feature space.
- Fisher
Information - Fisher information I(θ) = E[(∂/∂θ log p(x;θ))²].
- Gaussian
Complexity - Gaussian complexity (Gaussian analog of Rademacher).
- Gaussian
Process - Gaussian process regression model.
- Gradient
Boosting - Ensemble method: gradient boosting.
- Growth
Function - Growth function Π_H(m) for a hypothesis class.
- KLDivergence
- KL divergence D_KL(P‖Q) = Σ P(x) log(P(x)/Q(x)).
- Kernel
Function - A kernel function k: ℝ^d × ℝ^d → ℝ.
- Kernel
Matrix - Kernel (Gram) matrix K_{ij} = k(x_i, x_j).
- KernelSVM
- Kernel SVM dual representation.
- KernelSVM
Trainer - Kernel SVM trainer using a simplified SMO algorithm.
- Lasso
Reg - Lasso (ℓ₁) regularization: min_h L(h) + λ‖h‖₁.
- Mutual
Information - Mutual information I(X;Y) = H(X) + H(Y) - H(X,Y).
- Online
Gradient Descent - Online Gradient Descent with regret bound O(√T).
- PACBayes
Generalization - PAC-Bayes generalization bound computation.
- PACLearner
- PAC learner: wraps accuracy/confidence parameters.
- Perceptron
- Online Perceptron classifier.
- Rademacher
Complexity - Rademacher complexity estimate for a finite hypothesis class.
- Regret
Bound - Regret bound summary.
- SVMClassifier
- Support Vector Machine classifier.
- Sample
Complexity - Sample complexity for PAC learning (Blumer et al. / Vapnik-Chervonenkis bound).
- Tikhonov
Reg - Tikhonov (ridge) regularization: min_h L(h) + λ‖h‖².
- UCBBandit
- UCB1 (Upper Confidence Bound) algorithm for multi-armed bandits.
- Uniform
Convergence - Uniform convergence checker.
- VCDimension
- VC dimension and growth function calculations.
Enums§
- GBLoss
- Kernel
Type - Available kernel types.
- SVMKernel