Deep Learning (DL) · Beginner level

Beginner level · Module 4 of 5 · 11 guided hours

Transfer Learning with Pre-trained Models

Reuse models trained on large data sets to get strong results with small local data sets.

Learning objectives

  • Explain why transfer learning works
  • Use a pre-trained network as a feature extractor
  • Fine-tune a pre-trained network on a new task

Topics

  • pre-trained models and model hubs
  • feature extraction
  • replacing the classifier head
  • freezing and unfreezing layers
  • fine-tuning learning rates
  • licences of pre-trained models

Lessons

  1. Lesson 1 Why Transfer Learning Works

    Understand how features learned on large data sets can be reused for new tasks.

    pre-trained modelgeneral vs specific featuresmodel hublicence

  2. Lesson 2 Feature Extraction

    Use a frozen pre-trained network as a feature extractor and train a new classifier head on top.

    frozen layersclassifier headrequires_gradfast training

  3. Lesson 3 Fine-Tuning

    Unfreeze some pre-trained layers and train them gently to adapt the model further to your data.

    unfreezingsmall learning ratecatastrophic forgettingcomparison

Practical activity

Build the leaf disease classifier by fine-tuning a pre-trained CNN on a provided open data set of leaf photos.

Expected outcomes

  • Strong models from small data sets
  • Correct, licence-aware use of pre-trained models

Assessment

The level project's model and evaluation, assessed against its criteria.

Free taster lesson Transfer Learning & Pre-trained Models Read a free self-study lesson on this topic in the Free Course Library.

Taught live by an instructor — online or in person. Free course lessons are for self-study and do not lead to a certificate — see registration options.