Zheng Wang

Towards Autonomous Compiler Design Using Machine Learning

In recent years, machine learning has shown promise in making compilers more effective for code optimisation. A traditional human-derived compiler decision model can be replaced by a machine-learned version based on empirical observations from training benchmarks. Given the massive success of machine learning in domains like natural language processing and autonomous systems, this technology can fundamentally change the way compilers are designed and developed, allowing compilers to catch up with fast-evolving hardware to deliver scalable performance without needing years of compiler experts' time. However, many problems remained, limiting the scale on which machine learning in compilers can operate. 

In this talk, I will present some of my collaborative work to enable compiler developers to more easily integrate machine learning into compiler design. I will outline some of the challenges of integrating machine learning with compilers. 

back to overview

Watch Recording



Zheng Wang is professor of intelligent software technology at the School of Computing at the University of Leeds and a Turing Fellow at The Alan Turing Institute. He works at the intersection of machine learning and compilers and is known for his work in incorporating machine learning into compilation technology. He has published over 100 papers and received four best paper awards. His research has been successfully transferred into various industry settings.