Michael Defferrard

Title: Michael Defferrard and Machine Learning for Netlist Design.

Introduction

Michael Defferrard is an inventor associated with Chavannes-sous-Orsonnens, Switzerland. His patent work applies machine learning to the evaluation of electronic circuit designs.

Latest Patents

Defferrard is associated with 1 patent, titled “Machine Learning for Netlist Design.” The invention describes accessing a directed graph representing an electrical circuit’s netlist, selecting a node corresponding to a register, and generating a subgraph that captures the register’s connectivity cone. A trained machine-learning encoder then creates a functional embedding of the subgraph, which is used to predict a performance characteristic of the netlist design.

Career Highlights

Defferrard is associated with Qualcomm Incorporated. His patent reflects work at the intersection of machine learning and electronic circuit design, with a focus on predicting design performance from netlist structure.

Collaborations

Gokce Sarar is identified as a coworker. The available information does not specify their particular collaboration or contributions to the patent.

Conclusion

Michael Defferrard’s patent illustrates how machine learning can help evaluate electronic circuit designs by analyzing netlist connectivity and predicting performance.