A Benchmark Suite to Evaluate DNN’s Resilience
Assessing AI systems reliability is essential before deploying them in safety-critical applications. While recent efforts have focused on improving model resilience to random hardware faults, meaningful comparison remains difficult due to the lack of standardized reference models. Different authors use different implementations, which makes comparisons unfair and biased: resilience is influenced by the training processes, the software framework, and data representations. To address these issues, this work introduces a benchmark suite of CNN models to test the resilience of DNNs. The benchmark is structured on different axes: software framework, hardware platform, data representation, task and dataset. It is aimed at providing a shared foundation for fair and reproducible resilience evaluation.
Authors
- Luca Cassano (ORCID: https://orcid.org/0000-0003-3824-7714)
- Ernesto Sánchez (ORCID: https://orcid.org/0000-0002-7042-295X)
- Salvatore Pappalardo
- Cristiana Bolchini (ORCID: https://orcid.org/0000-0001-5065-7906)
- Antonio Miele (ORCID: https://orcid.org/0000-0003-3197-0723)
- M. Sonza Reorda (ORCID: https://orcid.org/0000-0003-2899-7669)
- V. Turco (ORCID: https://orcid.org/0009-0001-0754-8189)
- Alberto Bosio (ORCID: https://orcid.org/0000-0001-6116-7339)
- Annachiara Ruospo (ORCID: https://orcid.org/0000-0003-2040-9762)
- Dario Passariello
Institutions
- Université Claude Bernard Lyon 1 (FR)
- École Centrale de Lyon (FR)
- Centre National de la Recherche Scientifique (FR)
- Politecnico di Torino (IT)
- Institut des Nanotechnologies de Lyon (FR)
- Institut National des Sciences Appliquées de Lyon (FR)
- Politecnico di Milano (IT)
Publication Details
- Published
- 2025-09-20
- DOI
- https://doi.org/10.1109/itc58126.2025.00090
- Primary Topic
- Radiation Effects in Electronics
- Type
- article
- Field-Weighted Citation Impact
- 0.00