| KEYWORD |
Area Engineering
Image Test Libraries for the in-field test of neural networks-based systems
keywords ARTIFICIAL INTELLIGENCE, ARTIFICIAL NEURAL NETWORK, FAULT DETECTION, FAULT INJECTION, RELIABILITY, TESTING
Reference persons ANNACHIARA RUOSPO, EDGAR ERNESTO SANCHEZ SANCHEZ, MATTEO SONZA REORDA
Research Groups DAUIN - GR-05 - ELECTRONIC CAD and RELIABILITY GROUP - CAD
Description Over the past years, Artificial Neural Networks (ANNs) have become highly prominent in diverse artificial intelligence applications, proving remarkable performance across numerous tasks. Nonetheless, as ANNs find greater integration in safety-critical domains, the need to ensure their safety and reliability has emerged as a crucial issue. Efforts are underway to establish safety standards that specifically address the unique challenges posed by ANNs. As an example, these standards emphasize the importance of detecting in field permanent and transient faults that may occur during the operational phase of devices (e.g., GPUs, ASICs, hardware accelerators).
The main intent of the thesis is to implement innovative methodologies to turn high-quality test stimuli in the form of test images, to be executed during the normal inference process of a neural network. An Image Test Library (ITL) is composed of a set of carefully-developed test images that can be executed in the field by the targeted neural network model to test specific hardware units.
Deadline 27/11/2024
PROPONI LA TUA CANDIDATURA