Matthieu Martel

Papers

Total Citations

H-Index

About

Matthieu Martel is a leading researcher in the intersection of formal methods, embedded systems, and numerical computation, with a particular focus on fixed-point arithmetic and its application to neural networks. His most impactful work addresses the critical challenge of deploying neural networks in safety-critical, resource-constrained environments such as robots, rockets, and autonomous vehicles. Martel’s major contributions lie in developing automated code generation and synthesis techniques that convert floating-point neural network models into efficient, verified fixed-point implementations. His 2022 paper "Code Generation for Neural Networks Based on Fixed-point Arithmetic" (7 citations) pioneers methods to reduce the time and memory overhead of neural networks, making them compatible with embedded systems without sacrificing reliability. In subsequent works like "Fixed-Point Code Synthesis for Neural Networks" (4 citations) and "Fixed-Point Code Synthesis Based on Constraint Generation" (2 citations), he advances constraint-based approaches to ensure numerical accuracy and correctness. Martel’s research is pivotal for bridging the gap between high-performance AI and the stringent requirements of safety-critical applications, offering practical tools for engineers and inspiring further work in verified numerical computing.