PerFlex Project

Performant and Flexible digital Systems through Verifiable AI

Performant and Flexible digital Systems through Verifiable AI

A collaborative research project addressing performance optimization in multicore processor systems by combining machine learning with formal verification methods.

Status: Ongoing (September 2022 – August 2025)

Key Focus:

  • Developing transparent AI solutions that preserve non-functional properties like execution time and power consumption
  • Characterizing processor activities to maintain individual task performance
  • Creating verifiable AI approaches supporting deployment and system maintenance

Participating Organizations:

  • Mälardalen University (expertise in ML/AI, formal methods, verification, multicore systems)
  • Hitachi Energy AB (embedded and cyber-physical systems)
  • Ericsson AB (time-critical systems)

Key Personnel: Cristina Seceleanu, Ning Xiong, Tiberiu Seceleanu (project leader), and Peter Backeman

For more information, visit the PerFlex project page.