Theory of Intelligence
DEEP LEARNING•Computer Vision•Cnn Architectures•2.1 LeNet & AlexNet

Introduction

The two networks that started it all: LeNet-5, which read handwritten digits in the 1990s, and AlexNet, whose 2012 ImageNet win launched the deep learning era.

timer1 min read•format_list_bulletedConcept 1 of 3•forumLesson Discussion

Introduction

Every modern vision network is built from the same pieces you met in Chapter 01: convolutions, activations, pooling and a classifier at the end. Architectures differ in how they arrange those pieces. Studying the famous ones is like studying classic buildings in architecture school: each solved a real problem, and their ideas are reused everywhere.

This topic covers the two pioneers. LeNet-5 showed that a CNN trained with backpropagation could read handwriting reliably. AlexNet, twenty years later, showed that a much bigger CNN, trained on a GPU with lots of data, could beat every other method at recognizing everyday objects by a huge margin.

Discuss theoretical concepts, mathematical formulas, and architectural foundations.

Supports inline code with `backticks`
No comments yet. Be the first to start the discussion!
Next Lessonarrow_forward
2.2 VGG & GoogLeNet