Deep Learning Essentials

When to use fine-tuning

There are two major factors when considering whether to use fine-tuning—the size of the new dataset, and its similarity to the original dataset. Following are two cases where you can utilize fine-tuning:

• When the new data/domain is similar to the one used for training the existing model and your new dataset is small
• When your new dataset is quite different from the data used in the pre-trained model, but with enough samples, you may fine-tune the model even for a large number of layers but with a small learning rate