Lesson 18 of 21 · Part V, Deep learning and beyond

Randomized neural networks

Randomness as a resource: Random Forests, hidden layers with random weights that are never trained, a linear readout in one step, Cover’s theorem; pros and cons of random features.

4 figures to manipulate.

In this lesson

• Randomness in Machine Learning
• Neural networks with random weights

Other lessons

Deep learning · Unsupervised learning — K-means and SOM · All lessons

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