Artificial intelligence-based design automation algorithms tested in research studies show improved runtime and inference results over traditional EDA approaches. What does it take to move machine learning, deep learning, and reinforcement learning from research feasibility studies to production-worthy AI design automation tools?
For AI-based EDA tools, this panel discussion will explore the following:
• How do we feed the AI beast? What data is needed? Where does it come from and what relationships are required?
• What software platforms are available today to move research-level prototypes to production-level flows?
• Can research-level systems be successfully used in-house to improve hardware and software training and inference solutions? What criteria would necessitate a transition to production level?
• What I/O and data structure standards are needed to enable different AI engines for training and inference, shared AI models, and state-of-the-art visualization tools