Testing AI Models - Experiences, Techniques and Tools

  • 1 h 30 min

There is AI in the air. A part and parcel of modern life, the usage of AI/ML in everyday life is ever increasing. However, testing the AI/ML models remains an arcane art. From functional tests to various non-functional tests including data privacy leakage and adversarial attacks, various tests need to be conducted on different models. We will look at application of the Linguistic Analysis method and metamorphic testing, back-to-back testing etc. for testing of AI  models. In addition to these, we will look at explainability, coverage and performance-related tests. 

The techniques and case studies presented are based on personal experiences of testing AI/ML models and also building a product AIEnsured for testing AI models. Some of the content of this talk is derived from AiU, the world's first AI testing certification, now a part of the ISTQB CTAI.

Comments ({{Comments.length}} )
  • {{comment.AuthorFullName}}
    {{ comment.DateCreated | date: 'dd.MM.yyyy' }}

To leave a feedback you need to

Chat with us, we are online!