Learn Before
Comparison

Modular vs. End-to-End Driving Systems with Limited Paired Training Data

A modular self-driving pipeline learns separate perception and decision stages, such as detecting lanes, vehicles, and traffic signals before choosing an action. When labels for these intermediate tasks are more available than paired sensor-input and driving-action examples, the modular pipeline is usually more practical than a fully end-to-end model. As direct input-to-action training data increases, the end-to-end option becomes more feasible.

0

1

Updated 2026-08-30

Contributors are:

Who are from:

Tags

Machine Learning

Deep Learning

Supervised Learning

Dive into Deep Learning @ D2L

Data Science

Machine Learning Strategy

Machine Learning Yearning @ DeepLearning.AI