ManySpace
Türkçe Sign in Sign up

AI & MLRecommendedAIML-SYS-020

Retraining and Continual Learning in Production

When and how to update deployed models as data drifts, without forgetting or breaking downstream consumers.

Difficulty
3 / 5
Importance
4 / 5
Study time
1–2 h
Kind of work
Mixed

Why it matters

The learner can choose a retraining trigger and schedule for a drifting system and name the risk of each option.

The route to this topic

Route AIML-SYS-020
What Is a Model? Learning from Data Tasks: Regression, Classification, and Beyond Features and Targets Datasets and Splits Classification Framing Linear Regression I: Prediction and Residuals Embeddings Sampling and Splits Scores and Logits Loss and Fitting: Squared Error Distribution Shift The Artificial Neuron Linear Regression II: Least Squares Fitting Objective Functions and Empirical Risk Overfitting and Underfitting Activation Functions Gradient Descent in Machine Learning Training vs Serving The Multilayer Perceptron Monitoring and Drift Stochastic and Mini-batch Optimization Transfer Learning Fine-Tuning Catastrophic Forgetting Retraining and Continual Learning in Production