Import Questions

This page will contain a list of questions that I will continuously revisit to keep focused on answering the highest utility questions.

How do we certify that a system satisfies a given set of risk criteria?

  • What is risk?
  • What are risk criteria?
  • How do we certify ML-based systems achieve it?
  • How do we certify perception-based control systems achieve it?
  • How do we certify ML-based, perception-based control systems achieve it?

How can we test and iteratively improve our model to capture the relevant, dominant dynamics?

  • interested in abstract, general approaches to this question for use in improving engineering models and implicit/explicit mental models.