AI-based tools for the optimized design/sizing of integrated circuits (T4)

Objective: AI-based tools for the optimized design and sizing of radio-frequency (RF), analog, and mixed-signal integrated circuits, with applications in intelligent systems

Research challenges / Novelty / Innovation
Challenges: high sensitivity to process, temperature, and voltage variations, large number of required Monte Carlo simulations, high computational cost of algorithms, small training datasets

Research results:
• Methods and algorithms
• The research results will lead to the development of 4 (four) integrated circuits and at least 2 (two) invention/innovation applications

Inovation:
• Replacement of Monte Carlo simulations with multivariate adaptive ML models
• Integration of reinforcement learning (RL) with knowledge transfer
• Automation of layout generation and PEX simulations using AI