An Introduction to Drift Detection

Explore the fundamentals of drift detection in machine learning through this comprehensive 1 hour 29 minute workshop from the Toronto Machine Learning Series. Gain hands-on experience with Alibi Detect, an open-source Python library, to detect and analyze various types of drift in real-world scenarios. Learn to set up drift detectors, identify drift types, and preprocess different data forms for detection. Discover state-of-the-art detectors for fine-grained attribution and experiment with model uncertainty-based detectors. Delve into the novel context-aware drift detector to test for conditional drift. Suitable for beginners to intermediate learners with basic machine learning knowledge and Python experience, this workshop covers crucial topics including the importance of drift detection, anatomy of drift detectors, and practical applications in production environments. Work through Jupyter notebooks to explore graph thinking, graph algorithms, graph embeddings, and graph neural networks, equipping yourself with essential skills to maintain the performance of deployed machine learning models.

An Introduction to Drift Detection