MetaLight: Value-based meta-reinforcement learning for traffic signal control

Keyphrases

• Value-based 100%
• Traffic Signal Control 100%
• Reinforcement Learning Model 50%
• Adaptation 25%
• Learning Algorithm 25%
• Learning Process 25%
• Global Level 25%
• Training Data 25%
• Gradient-based 25%
• Computational Resources 25%
• Transportation Problem 25%
• Reinforcement Learning 25%
• Reinforcement Learning Algorithm 25%
• Traffic Jam 25%
• New Scenario 25%
• Adaptive Rate 25%
• Meta-learning 25%
• Traffic Accidents 25%
• Traffic Scenarios 25%

Computer Science

• Reinforcement Learning 100%
• Traffic Signal 100%
• Speed-up 20%
• Learning Algorithm 20%
• Structure Model 20%
• Individual Level 20%
• Training Data 20%
• Transportation Problem 20%
• Learning Process 20%
• Computational Resource 20%