VVZ API is not affiliated with ETH Zurich. Data might be outdated or incorrect. Please view the official ETHZ Vorlesungsverzeichnis for binding information.
Abstract
The course guides participants in teams through building end-to-end ML systems for real business problems. Covering the full lifecycle from problem formulation to deployment, participants tackle real-world challenges: imperfect data, bias/fairness, regulatory compliance, and performance trade-offs. Hands-on work includes a baseline pipeline plus optional extensions in areas of interest.
Objective
(1) design and implement a complete ML pipeline from problem formulation to API deployment (2) identify, implement and evaluate suitable models for a given task (3) evaluate and mitigate bias, fairness, and regulatory risks (4) make and defend architectural decisions based on real-world constraints
Content
Main topics: problem formulation; data preparation; algorithm selection & feature engineering; hyperparameter optimization; model evaluation & testing; bias, fairness & regulatory compliance; interpretability methods. Additional extensions available: advanced architectures, task-specific fairness metrics, visualizations, API deployment, and documenting experimental failures.
Resources
Lecture Notes
Slides and links to extra material will be distributed during the course.