Summary  
The chapter describes different categories of transfer learning (inductive, transductive, unsupervised) based on how source and target domains and tasks relate, and explains how learned features can be reused across tasks.  

General domain of usage  
Machine learning

## Types of Transfer Learning

Transfer learning can be categorized based on the relationship between source and target domains and tasks:

**1. Inductive Transfer Learning:**
- The target task is different from the source task;
- **Example**: using an image classifier trained on animals to classify medical images.

**2. Transductive Transfer Learning:**
- The source and target tasks are the same, but the domains differ;
- **Example**: sentiment analysis on English reviews (source) and French reviews (target).

**3. Unsupervised Transfer Learning:**
- Both tasks are unsupervised (e.g., clustering), but domains differ.

**Notation:**
$$
\begin{aligned}
D_S &\neq D_T \quad &\text{(domains differ)} \\
T_S &\neq T_T \quad &\text{(tasks differ)}
\end{aligned}
$$

**Feature reuse**: often, early layers of a neural network learn general features (e.g., edges in images) that can be reused across tasks.

The choice of transfer learning type depends on how similar the source and target domains and tasks are.

Note

Which statements correctly describe the types of transfer learning?

Master the core concepts and hands-on techniques of transfer learning. Learn how to leverage pre-trained models for image classification and sentiment analysis, and gain practical experience with CNNs and transformers.

Explore the core principles, terminology, and motivations behind transfer learning. Build a strong theoretical base before diving into practical applications.

Apply transfer learning to image classification using pre-trained CNNs. Learn how to fine-tune models like VGG16 for new tasks.

Explore transfer learning for NLP using transformer models. Learn to adapt BERT for sentiment analysis.