Action evaluation hardware accelerator for next-generation real-time reinforcement learning in emerging IoT systems

Keyphrases

• Hardware Accelerator 100%
• Internet of Things System 100%
• Interactive Reinforcement Learning 100%
• Action Evaluation 100%
• Post-decision State 100%
• State-based 66%
• Decision Policy 66%
• Communication Channels 33%
• Optimal Decision 33%
• Unique Characteristics 33%
• Computationally Intensive 33%
• Application Domain 33%
• Learning Framework 33%
• Dynamic Computation 33%
• Software Implementation 33%
• Reinforcement Learning 33%
• Learning Rate 33%
• Processing Load 33%
• Internet of Things Sensors 33%
• Q-learning 33%
• Reinforcement Learning Framework 33%
• Latency-sensitive 33%
• State Value 33%
• Source Dynamics 33%
• State Learning 33%
• Efficient Hardware Architecture 33%
• Unknown Dynamic Environment 33%
• Value Expectation 33%
• Unknown Statistics 33%

Computer Science

• Internet of Things 100%
• Hardware Accelerator 100%
• Reinforcement Learning 100%
• Postdecision State 100%
• Learning Framework 66%
• Decision Policy 66%
• Computational Cost 33%
• Speed-up 33%
• Hardware Architecture 33%
• Application Domain 33%
• Data Source 33%
• Software Implementation 33%
• Dynamic Environment 33%
• Priori Knowledge 33%