AI & Functional Safety: Deploying Neural Networks on Safety Hardware — Quantization and Inference Integrity
- Date
- 2027-06-07
- Location
- Online
- Host
- Zach L
About this event
A live 30-minute expert session on Deploying Neural Networks on Safety Hardware — Quantization and Inference Integrity (AI / ML — AI & Machine Learning Safety). What We'll Cover: Where deployment transformations sit in the lifecycle and why the deployed model, not the trained one, is the safety item Step by step: quantization-aware training or post-training quantization, equivalence testing, accelerator fault modes, redundancy options How inference-path integrity maps to hardware metrics thinking and the model equivalence report it produces Common mistakes: verifying float32 offline and shipping int8 on-target with no delta characterization The bit-exactness or bounded-difference evidence and on-target test log an auditor cross-checks Related topics: model quantization · int8 inference · e