Deep Reinforcement Learning for Analog Circuit Sizing with an Electrical Design Space and Sparse Rewards.
Yannick UhlmannMichael EssichLennart BramlageJürgen ScheibleCristóbal CurioPublished in: MLCAD (2022)
Keyphrases
- design space
- reinforcement learning
- analog circuits
- fault diagnosis
- markov decision processes
- state space
- design choices
- design process
- design space exploration
- design tools
- response surface
- digital circuits
- wavelet packet transform
- search space
- reward function
- reinforcement learning algorithms
- neural network
- learning algorithm
- sparse representation
- learning process
- multiscale
- artificial intelligence
- machine learning
- control policy
- reward shaping
- online learning
- low cost
- search algorithm
- optimal solution
- knowledge base
- real world