> ML_ALGORITHM // TREE-STRUCTURED-PARZEN-ESTIMATORS-TPE_v1.0
Tree-structured Parzen Estimators (TPE)
Scalable Bayesian optimization algorithm modeling parameter densities of top-performing trials separately from bottom trials to maximize expected improvement.
Bayesian Optimizationevolutionary-searchhigh-intrinsicsmall (<1k)
Back to All AlgorithmsComputational Complexity
Training Complexity:O(trials * log(trials))
Inference Complexity:O(candidates * kernel_eval)
Hardware Profile
CPU Friendly:Yes
Requires GPU:No
Memory Footprint:low
Interpretability & Data
Interpretability Tier:high-intrinsic
Training Data Needs:small (<1k)
Interpretability Assessment
Directly displays which hyperparameter ranges yield top quantile performance via 1D kernel densities.
Suitable Tasks & Supported Modalities
Suitable Tasks:
hyperparameter tuningmixed discrete continuous hpo
Supported Modalities:
tabular
Implementing Libraries
Foundational Literature
Common Pitfalls & Warnings
- Assumes tree independence between hyperparameters, failing to capture subtle multi-parameter diagonal correlations