Thuật ngữ AI
Từ điển đầy đủ về Trí tuệ nhân tạo
Covariance Function
Kernel function defining similarity between points in a Gaussian process, determining the correlation structure and smoothing properties of the model.
GP-MDP
Markov decision process where rewards or transitions are modeled by Gaussian processes, integrating uncertainty into the RL problem formulation.
Thompson Sampling with GP
Action sampling method based on draws from the Gaussian process posterior, enabling exploration guided by uncertainty.
GP-SARSA
Variant of the SARSA algorithm using Gaussian processes to model the Q-value function with uncertainty in reinforcement learning.
RBF Kernel
Gaussian radial basis function used as covariance in GPs, ensuring continuity and differentiability of modeled functions.
GPQ
Q-function approximated by Gaussian process, allowing quantification of uncertainty on value estimates in reinforcement learning.
GP-PILCO
RL framework using Gaussian processes to learn system dynamics with analytical propagation of uncertainty.
Contextual Bandit with GP
Bandit problem where rewards depend on a context, modeled by Gaussian processes to capture spatial uncertainty.
GP-MCTS
Monte Carlo Tree Search utilisant des processus gaussiens pour évaluer les nœuds avec incertitude, améliorant la recherche dans RL.
GP-MAE
Maximum a Posteriori avec estimation de l'incertitude par processus gaussiens pour l'estimation de politique en apprentissage par renforcement.