DE-TW-TEdgeAI
Design Concepts and Methods for Trustworthy and Attack-Resistant AI Chips in Edge Computing
DE-TW-TEdgeAI develops design concepts and methods for trustworthy and attack-resistant AI chips in edge computing environments. As edge AI becomes more prevalent, the demand for hardware resistant to adversarial attacks increases.
Objectives and Approach
The goal is to develop AI accelerators for edge systems based on neural networks, offering significant efficiency advantages while incorporating trustworthiness and hardware security. The project implements open-source tools to generate optimized hardware descriptions from AI models and extends existing design tools with security analysis methods for neural networks.
Innovations and Perspectives
The project strengthens German-Taiwanese collaboration in microelectronics and expands the scientific network for developing chip design talent. It enhances innovation capacity at German research institutions by supplementing the open-source tool landscape with appropriate analysis and hardening methods for AI networks.
Project Coordinator
- Universität zu Lübeck
Project Partners
- National Taiwan University, Taipei
- Chung Yuan Christian University, Taoyuan
This project is funded by the German Federal Ministry for Research, Technology and Space (BMFTR) as part of the Design Initiative Microelectronics.
