Learning from your (model’s) mistakes

Short talk

Go to NumFOCUS academy page.

Understanding in what situations an ML model fails is essential to attempt to fix the model’s flaws. For data scientists concerned with robust ML system design, this talk will show how to streamline error analysis thanks to Model Performance Predictors, by automatically breaking down model failures into meaningful clusters and comparing them with the successfully predicted baseline.

Speaker

Simona Maggio

After obtaining a PhD in Biomedical Image Processing in 2011, Simona Maggio worked in several companies (CEA, Thales, Rakuten) as a Research Engineer in Computer Vision and Natural Language Processing for applications ranging from video surveillance to document digitization and e-commerce. She’s now Senior Research Scientist at Dataiku, exploring MLOps topics, such as model debugging, robustness and interpretability.