Glosarium AI
Kamus lengkap Kecerdasan Buatan
Batch Inverse Reinforcement Learning
IRL method that learns from a fixed set of expert demonstrations without active interaction with the environment.
Active Inverse Reinforcement Learning
IRL approach where the agent actively selects actions to execute to better understand the expert's reward function.
Bayesian Inverse Reinforcement Learning
IRL framework using Bayesian inference to model uncertainty about the reward function from demonstrations.
Inverse Reinforcement Learning from Preferences
IRL method that infers rewards from comparisons of preferences between trajectories rather than complete demonstrations.
Hierarchical Inverse Reinforcement Learning
An IRL approach that decomposes complex tasks into hierarchical subtasks to learn multi-level reward functions.
Multi-Agent Inverse Reinforcement Learning
Extension of IRL to environments where multiple agents interact and must learn collective or individual rewards.
Deep Inverse Reinforcement Learning
Use of deep neural networks to represent and learn complex reward functions from high-dimensional data.
Adversarial Inverse Reinforcement Learning
IRL framework using adversarial techniques where the generator and discriminator compete to learn the reward.
Inverse Reinforcement Learning with Reinforcement Learning
Method transforming the IRL problem into a standard RL problem where the agent learns to maximize the likelihood of demonstrations.
Semi-Supervised Inverse Reinforcement Learning
IRL approach combining labeled demonstrations with unlabeled data to improve reward learning.
Inverse Reinforcement Learning for Robotics
Specialized application of IRL for learning robotic behaviors from human demonstrations in manipulation and navigation.
Inverse Reinforcement Learning with User Feedback
IRL method actively integrating qualitative user feedback to iteratively refine the reward function.
Cooperative Inverse Reinforcement Learning
IRL framework where humans and AI actively collaborate to jointly define and optimize reward objectives.
Inverse Reinforcement Learning for Planning
Using IRL to extract implicit goals from existing plans to improve future planning systems.
Inverse Reinforcement Learning by Maximum Entropy
IRL approach favoring reward solutions with maximum entropy to avoid overfitting to demonstrations.