Overview
This workshop aims to address emerging challenges and explore innovative solutions in the field of computer-aided design (CAD) for integrated circuits and systems for ultra low latency and high bandwidth scientific applications. The workshop builds on the ideas laid out in the "Applications and Techniques for Fast Machine Learning in Science" white paper and the corresponding Fast Machine Learning for Science conference series (2023 edition). This workshop at ICCAD 2023 aims to bring domains together and forge new connections with the CAD community.
Scientific applications across particle physics, astrophysics, material sciences, quantum information sciences, fusion energy (and beyond!) utilize data acquisition and in situ processing systems which require very low latency and high data bandwidth custom processing elements and real-time control modules. Integrating data reduction and control applications with real-time machine learning algorithms can enable significant breakthroughs in the sciences. We will bring together researchers, practitioners, and industry experts to exchange ideas, share applications, and discuss the latest advancements in CAD methodologies, algorithms, and tools.
Topics of interest
The areas of interest related to real-time scientific applications include but are not limited to:
- Methods and tools for efficient algorithm design, implementation, and integration methodologies
- Software-hardware codesign, partitioning, and optimizations
- Design automation and synthesis, timing analysis
- Physical design and layout
- High-level synthesis and system-level design for edge AI hardware
- Robust machine learning, anomaly detection, and fault tolerance
- Continuous, adaptive, and reinforcement learning for low latency control
- Emerging technologies in CAD machine learning and AI-assisted design
© Fast Machine Learning for Science @ ICCAD, 2023