AI-ordlista
Den kompletta ordlistan över AI
SMAC
Bayesian optimization tool using random forests as a surrogate model for algorithm configuration, particularly effective in categorical and conditional hyperparameter spaces.
Ensemble Selection
Process of automatically building an ensemble of optimized models by dynamically selecting and weighting the best models from a large pool of candidates.
Optimized Pipeline
Complete sequence of data transformations and machine learning models automatically optimized to maximize predictive performance on a given dataset.
Bootstrap Ensembling
Ensemble technique where multiple models are trained on different bootstrap samples of the training dataset to reduce variance and improve generalization.
Algorithm Configuration
Process of systematically searching for the best hyperparameter configuration for a given algorithm on a specific class of problems.
Conditional Search Space
Hyperparameter space where the validity of certain parameters depends on the values of other parameters, requiring adaptive search strategies.
Auto-sklearn 2.0
Improved version of Auto-sklearn with advanced parallelization mechanisms, meta-learning sampling strategies, and more efficient ensemble integration.
Meta-model
Model trained to predict the performance of machine learning algorithms based on dataset meta-features to guide algorithm search.
Performance-based Model Selection
Automatic model selection strategy based on comparative evaluation of multiple configurations' performance on validation data.
Resource-aware Optimization
Optimization process that takes into account computational constraints such as time and memory to find the best performance-cost tradeoff.