What is Transfer Learning?
Transfer Learning — Taking an AI model trained on one task and repurposing it as the starting point for a different but related task.
Transfer learning takes knowledge a model learned on one task and applies it to a different but related task. Instead of training from scratch, you start with a pre-trained model and adapt it to your domain. This dramatically reduces training time, data requirements, and costs.
Frequently Asked Questions
Is fine-tuning the same as transfer learning?
Fine-tuning is the most common form of transfer learning. The broader concept also includes feature extraction (using a frozen pre-trained model as a feature generator) and domain adaptation.
Why does transfer learning work?
AI models learn general features in early layers (edges, textures, grammar) that are useful across many tasks. Only the later, task-specific layers need retraining for new applications.
When does transfer learning fail?
When the source and target domains are too different. A model trained on English text transfers well to legal English but poorly to medical imaging. Domain similarity is key.