Yassamine Seladji

Papers

Total Citations

H-Index

About

Yassamine Seladji’s research lies at the critical intersection of neural network deployment and embedded systems, focusing on making deep learning viable for safety-critical applications like autonomous driving, robotics, and aerospace. Her major contributions center on code generation and synthesis for neural networks using fixed-point arithmetic, addressing the fundamental challenge that floating-point computations are too memory- and time-intensive for resource-constrained embedded devices. Her most cited work, “Code Generation for Neural Networks Based on Fixed-point Arithmetic” (2022, 7 citations), demonstrates how to automatically translate neural network models into efficient fixed-point code, preserving accuracy while drastically reducing computational overhead. In her follow-up paper, “Fixed-Point Code Synthesis for Neural Networks” (2022, 4 citations), she further refines this synthesis process, enabling reliable deployment in systems where computational resources are limited but decisions must be made in real time. Seladji’s work is particularly notable for bridging the gap between high-performance neural network training and practical, low-power inference—a key enabler for next-generation autonomous systems. Her research has direct implications for making AI safer and more accessible in embedded environments, positioning her as an emerging voice in the field of trustworthy, efficient neural network implementation.