World Model as a Graph: Learning Latent Landmarks for Planning

Keyphrases

• World Model 100%
• Graph Learning 100%
• Deep Reinforcement Learning (deep RL) 40%
• Reachability 40%
• Model-based Reinforcement Learning 40%
• Novel Algorithm 20%
• Control Task 20%
• Robotic Manipulation 20%
• Human Intelligence 20%
• Learning Paradigm 20%
• Reinforcement Learning 20%
• Framework Model 20%
• Graph Search 20%
• Planning Horizon 20%
• Model Robustness 20%
• Q-function 20%
• Continuous Control Tasks 20%
• Multistep Transition 20%
• Long-horizon Planning 20%

Computer Science

• Deep Reinforcement Learning 100%
• Planning Horizon 100%
• Reinforcement Learning 50%
• Complex Environment 50%
• Model-Based Reinforcement Learning 50%
• Searching Algorithm 50%
• Model-Free Reinforcement Learning 50%
• Incorporate Planning 50%
• Continuous Control 50%