Jul
1
Paper Accepted at DAIES 2026: A Framework for the Automatic Deployment of Dependability Mechanisms on PyTorch-Based Models
July 1, 2026 | | Comments Off on Paper Accepted at DAIES 2026: A Framework for the Automatic Deployment of Dependability Mechanisms on PyTorch-Based Models
We are pleased to announce that our paper “A Framework for the Automatic Deployment of Dependability Mechanisms on PyTorch-Based Models” has been accepted for presentation at the 1st International Workshop on Dependable AI in Embedded Systems (DAIES 2026).
As Artificial Intelligence becomes increasingly integrated into safety- and mission-critical systems, ensuring the dependability of AI models has become a fundamental challenge. While numerous fault-tolerance techniques have been proposed, integrating them into existing neural networks often requires significant manual effort and expert knowledge.
In this paper, we present a framework that automates the deployment of dependability mechanisms in PyTorch-based neural networks. The proposed solution enables AI models to be enhanced with reliability features while minimizing developer intervention, facilitating the adoption of dependable AI practices throughout the development process.
The framework has been designed to support the systematic evaluation and integration of dependability mechanisms, helping researchers and practitioners improve the robustness of AI systems without modifying the original application logic. By reducing the complexity of incorporating fault-tolerance techniques, it contributes to accelerating the development of trustworthy AI solutions for safety-critical embedded systems.
We are proud to contribute to DAIES 2026, a forum dedicated to advancing the reliability, safety, security, and certification of intelligent embedded and autonomous systems.
We look forward to presenting our work and discussing future directions in dependable AI with the DAIES community.
More info at: https://safecomp2026.webs.upv.es/daies-2026
Comments