Training Models
summary
The talk covers methods for training and developing machine learning models, focusing on practical aspects of model upbringing (training, tuning and lifecycle) in Big Data and high-load environments. It discusses tools and practices used in Data Science, Machine Learning and AI to build and maintain models at scale for production systems.
AI-written summary · about
chapters (14)
- 0:00Model training & transfer learning
- 2:31Multimodeling & mixture of experts
- 6:45Teacher-student learning
- 10:38Hypernetworks & domain adaptation
- 13:24Model structure & parameter sharing
- 16:38Optimization & training schemes
- 20:51Neural networks as multimodels
- 27:05When multimodels are needed
- 33:52Mixture of experts simplified
- 36:59Q&A
- 40:31Ensemble learning limits
- 43:44Error & variance analysis
- 46:58Model selection advice
- 48:20Advanced learning resources
related talks
39mMachine Learning for Developers
41mWant a Good Machine Learning Model? Start with Data Processing
06mMachine learning is about your data and deployment
18mContinuous deployment of Machine Learning models
52mThe ML Lifecycle in Combat Conditions
33mAI Model Distribution Challenges and Best Practices
57mGood Practices for Applied Machine Learning
27mA Look Inside Applied Machine Learning