Martin Ferianc
Papers
1
Total Citations
3
H-Index
1
About
Martin Ferianc is a researcher whose work sits at the intersection of Bayesian deep learning and efficient neural network deployment, with a particular focus on uncertainty quantification under computational constraints. His most-cited paper, "On the Effects of Quantisation on Model Uncertainty in Bayesian Neural Networks" (2021), makes a foundational contribution by systematically investigating how model compression techniques—specifically quantisation—impact the reliability of uncertainty estimates in Bayesian neural networks (BNNs). This work is critical for deploying trustworthy AI in resource-limited settings, such as edge devices or autonomous systems, where both efficiency and calibrated confidence are paramount. Ferianc’s research addresses a key tension: while BNNs excel at providing uncertainty measures essential for safe decision-making, their practical adoption is hindered by high computational costs. By revealing how quantisation alters predictive uncertainty, he provides guidelines for building compact yet reliable models. With over 3 citations on this core paper, his work is gaining traction among researchers bridging Bayesian methods and model compression. Ferianc’s contributions are particularly valuable for students and practitioners seeking to deploy uncertainty-aware neural networks in real-world applications without sacrificing performance or safety.
Research Focus
Key Achievements
Top Papers
- 1