Chain of Assumptions in Supervised Statistical Learning
A supervised learning system is expected to perform well only when a chain of four assumptions holds, each verified on a different dataset:
- Low bias: the model fits the training set well on the cost function (e.g., comparable to human-level performance).
- Low variance: the model fits the dev (validation) set well on the cost function.
- The model fits the test set well on the cost function.
- The model performs well in the real world (e.g., happy users).
Because the checks are sequential, a failure at a later step points to the stage that needs attention (e.g., high dev-set error indicates a variance problem).
0
1
Contributors are:
Who are from:
Tags
Data Science
Related
Which one shows a basic machine learning workflow?
Classification Problem Representation
Classification Problem Evaluation
Classification Problem Optimization
Chain of Assumptions in Supervised Statistical Learning
A few useful things to know about machine learning
Common errors in machine learning
Chain of Assumptions in Supervised Statistical Learning
Understanding the Bias-Variance Tradeoff
Relationship between Capacity, Training and Test Errors, Bias-Variance Tradeoff in Machine learning
What is the issue with higher-order polynomials (high model complexity) in regard to fitting the training data and test data?
Under which circumstances will getting more training data help a learning algorithm to perform better?