Navigation: <-- Part VII: Unsupervised Learning | Main Index | Part IX: Closing the Loop - Projects in Practice -->

Part VIII: Deep Learning

Shallow models (linear models, trees, ensembles) cover a wide range of structured-data problems but hit a hard wall on raw images, audio, and sequences. This part explains why, and builds a conceptual map of deep learning. The goal is not a practical lab course but the vocabulary and intuition to use deep learning purposefully and recognize when it is and is not warranted.

Nuggets in This Part

#	Nugget	Prerequisites
1	When Shallow Models Fail	Start Simple
2	Building Blocks of Deep Networks	When Shallow Models Fail · Gradient Descent
3	Convolutional Neural Networks (CNNs)	Building Blocks of Deep Networks · Classification Tasks
4	What Deep Networks Learn: Representations	Building Blocks of Deep Networks · Convolutional Neural Networks (CNNs)
5	Transfer Learning	What Deep Networks Learn: Representations · Convolutional Neural Networks (CNNs)
6	Autoencoders	What Deep Networks Learn: Representations · Anomaly Detection
7	Transformers	Building Blocks of Deep Networks · Convolutional Neural Networks (CNNs)
8	Deep Learning in Practice: Choosing and Applying	Convolutional Neural Networks (CNNs) · Transfer Learning · Autoencoders · Start Simple

Script v1.7 (2026-07-28) · FGN