AI 용어집
인공지능 완전 사전
Stochastic Markov Decision Processes
MDP where transitions and rewards follow probabilistic distributions, modeling environmental uncertainty.
Monte Carlo Methods in RL
Algorithms using repeated random sampling to estimate state-action values in stochastic environments.
Stochastic Policies
Strategies returning probability distributions over actions rather than deterministic actions.
Bayesian Reinforcement Learning
Approach handling uncertainty over model parameters using probability distributions.
Multi-armed Stochastic Bandits
Exploration-exploitation problem where each arm has an unknown stochastic reward distribution.
Bootstrap Methods in RL
Techniques using resampling to quantify uncertainty in value estimates.
Gaussian Processes for RL
Using Gaussian processes to model uncertainty in the value or transition function.
Ensemble Methods in Stochastic RL
Combination of multiple estimators to capture epistemic uncertainty in learning.
Distributional Reinforcement Learning
Learning the full distribution of returns rather than only their expected value.
Quantile Regression DRL
Specific approach of distributional RL using quantile regression to model uncertainty.
Partially Observable Stochastic MDPs
Extension of stochastic MDPs with partial observation, increasing uncertainty about the state.
Stochastic Optimization in RL
Optimization methods accounting for noise and uncertainty in gradients and updates.