Glossary term
Glossary term
Foundations
Producing a model with poor predictive ability because the model hasn't fully captured the complexity of the training data. Many problems can cause underfitting, including:
Training on the wrong set of features.
Training for too few epochs or at too low a learning rate.
Training with too high a regularization rate.
Providing too few hidden layers in a deep neural network.
See Overfitting in Machine Learning Crash Course for more information.
Created for this library
A research team flags underfitting when both training and validation loss remain high and increases model capacity in response.
An ML team checks for underfitting before adding regularization because a model that already underfits should not be regularized further.
A retail forecasting team diagnoses underfitting by comparing training and validation loss and increases model complexity when both are high.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License