Cultivating a company-wide machine learning culture
summary
3M shows concrete wins: CloudMaker uses factory sensor data to predict web breaks; a polymer plant runs an RNN on S3/EC2 with SageMaker (TensorFlow/PyTorch) to predict 12‑hour reactor temperatures and free scheduling capacity; a Filtrate Filter with IoT sensors improves air quality and repeat purchases; unsupervised ML produces aligner treatment plans in one second versus hours; adversarial nets design metamaterial films; a model hub and scientist social network (Textract, Neptune) democratize reuse; growth comes from an annual hackathon of 800 people across 19 countries.
AI-written summary · about
chapters (9)
- 0:00Introduction to ML culture
- 0:46McKnight principles and innovation
- 5:08ML in factory operations
- 9:35Expanding ML to R&D
- 15:48Democratizing ML capabilities
- 17:18Materials informatics roadmap
- 22:35Scientist social network
- 26:17Hackathons and collaboration
- 29:41Culture and future vision
related talks
24mMachine Intelligence gets real
32mAutomating Production Production Level Machine Learning on AWS
19mSustaining ML in the enterprise
20mIndustrial-Scale Machine Learning
22mReal-world applications of machine learning: Beyond the hype
27mCulture & Organization for Effective ML at Scale
21mMastering the DevOps Machine Learning Pipeline
27mHow to make ML teams 80% more productive.