Deep Learning for Computer Vision

Tackling the underfitting and overfitting scenarios

The model may be sometimes too big or too small for the problem. This could be classified as underfitting or overfitting, respectively. Underfitting happens when the model is too small and can be measured when training accuracy is less. Overfitting happens when the model is too big and there is a large gap between training and testing accuracies. Underfitting can be solved by the following methods:

• Getting more data
• Trying out a bigger model
• If the data is small, try transfer learning techniques or do data augmentation

Overfitting can be solved by the following methods:

• Regularizing using techniques such as dropout and batch normalization
• Augmenting the dataset

Always watch out for loss. ...