Machine learning is about your data and deployment
summary
Argues that production matters more than fancy models: focus on three Ds — data cleaning, model development, and deployment — with deployment causing the hardest problems like retraining without downtime, feedback loops, model versioning, and wedging ML into COBOL and rules engines. Notes that classic algorithms often match deep nets, emphasizes model explanation for regulators and clinical safety, and points to IBM’s tooling (ibm.com/data-science, Watson, local ML hub) as practical solutions.
AI-written summary · about
chapters (3)
related talks
52mThe ML Lifecycle in Combat Conditions
26mWhy Machine Learning Algorithms Fall Short...
21mHow The Application Of Machine Learning Is Changing The Way Companies Use Data
27mA Look Inside Applied Machine Learning
33mAI with a DevOps mindset - experimentation, sharing ..
30mChallenges of machine learning: The new ML lifecycle
14mGetting Machine Learning applications into production using MLOps and CD4ML
47mMachine Learning in Production: An Intro to MLOps.