› Functional safety for artificial intelligence ‹
The main goal of functional safety is to ensure that automatic safety function will work correctly as it was designed or if is is impossible to perform in expected way then the system will fail in a safe manner.
In industry two main standard are defined: IEC61508 and IEC61511. These standards help to develop the safety lifecycle to eliminate unacceptable risk. However, these standards currently do not provide any guidance how to build safe systems, which use artificial intelligence (AI) algorithms. Therefore, our company try to bridge this gap and offer our experience and knowledge how to build safe products which use AI algorithms.
AI models validation and testing
We have the experience to guide you in providing robustness, interpretability, completeness and correctness of your AI models.
Training datasets specification
We can help you to verify your datasets wrt. to completeness and correctness.
Safety metrics
We know solutions that can help you to evaluate true reliability of your AI models.
Compliance to Standard
We have the knowledge how to weave your AI solutions into the work products of the official Standards.
Hazard Analysis for ML
We can guide you in finding ways to minimize the chance of hazardous behavior of your system.
Architecture for machine learning models
We can help you to establish a precise architecture of your AI solutions.