09-26, 10:05–10:55 (Europe/Amsterdam), Nebula
Not all mistakes in machine learning are equal—a false negative in fraud detection or medical diagnosis can be far costlier than a false positive. Cost-sensitive learning helps navigate these trade-offs by incorporating error costs into the training process, leading to smarter decision-making. This talk introduces Empulse, an open-source Python package that brings cost-sensitive learning into scikit-learn. Attendees will learn why standard models fall short in cost-sensitive scenarios and how to build better classifiers with Scikit-Learn and Empulse.
The confusion matrix doesn’t tell the whole story. In many real-world applications—fraud detection, medical diagnosis, churn, or loan approvals—the cost of a false positive can be vastly different from a false negative. Yet, traditional machine learning models optimize for overall accuracy, often ignoring these asymmetric costs and leading to poor decisions.
This talk will help attendees find their way through cost-sensitive learning concepts and introduce Empulse, a powerful yet easy-to-use package that brings cost-sensitive techniques to the scikit-learn ecosystem. Empulse also homogenizes a range of fragmented cost-sensitive methods—previously scattered across research papers and ad hoc implementations—into a consistent, scikit-learn-compatible API. Through live examples and real-world applications, participants will learn how to integrate cost-sensitive learning into their ML workflows and avoid the pitfalls of cost-insensitive models.
Who should attend? Data scientists, ML engineers, and researchers who want to improve model decision-making in cost-sensitive environments. Basic knowledge of scikit-learn and machine learning concepts (e.g., classification, evaluation metrics) will be helpful but not required.
Talk outline
- Why cost-sensitive learning? (5 min) – Why standard machine learning models fail in cost-sensitive scenarios
- Cost-sensitive learning fundamentals (10 min) – Cost matrices, cost-sensitive metrics, models and preprocessing techniques
- Implementing Cost-sensitive techniques (15 min) – Showing what you can already do in sklearn and how enhance it with Empulse.
- Q&A and discussion