Lesson 17 of 21 · Part V, Deep learning and beyond

Deep learning

Why many layers: hierarchical abstraction, no-flattening and compositionality; representation learning, autoencoders and transfer learning; distributed representations; the techniques: ReLU, clipping, batch normalization, dropout.

15 figures to manipulate.

In this lesson

• What deep learning is
• Are many layers really needed?
• Representation learning
• Distributed representations
• Other insights and recent results
• Techniques
• Applications, challenges and conclusions

Other lessons

Convolutional neural networks (CNN) · Randomized neural networks · All lessons

Notes by Fabio Piscitelli — Machine Learning (654AA), Prof. Alessio Micheli, University of Pisa, a.y. 2026/27.