Transformer‑Based Surrogate Modeling for Fast Electric Circuit Performance Prediction (Transformer‑Based Surrogate Model Module for Electric Circuit Performance Modeling, Tech ID: 24‑019)
Technology Overview: This technology applies transformer‑based machine learning models as surrogate predictors for electric circuit performance. By learning complex relationships between circuit configurations and performance metrics, the system enables rapid, near‑instant prediction without computationally expensive simulations. This approach significantly accelerates design iteration and optimization in electronic design automation workflows.
Industry Pain Point: Physics‑based circuit simulations are computationally intensive and slow, limiting design exploration.
NJIT Solution: Transformer‑based surrogate models provide fast, high‑accuracy performance prediction, enabling rapid iteration.
Key Features & Advantages
- Orders‑of‑magnitude faster than full simulations
- High prediction accuracy
- Integrates into EDA workflows
- Supports optimization and design exploration
Development Stage: TRL 3–4 – Model validation and benchmarking completed.
Target Markets
- EDA tool providers
- Integrated circuit designers
- AI‑driven hardware startups
Market Opportunity
- Global AI in EDA market (2026): ~$4B
- CAGR: ~25–28%
- Projected market size (2035): ~$35–40B
Commercial & IP Details
Inventors: Ningyuan Cao, Shaoze Fan, Chuang Gan, Xiaoxiao Guo, Jing Li, Shun Zhang, Xin Zhang