Essay

Analyze the connection between function simplicity and data requirements in an autonomous driving pipeline.

Question: Explain why breaking down autonomous driving into three subtasks (detecting cars, detecting pedestrians, and planning a path) allows the system to be trained with less data overall compared to a purely end-to-end approach. Focus on the complexity of the functions being learned.

Sample answer: Decomposing the autonomous driving pipeline into three distinct subtasks—detecting other cars, detecting pedestrians, and planning a path—reduces the complexity of the function that each individual model needs to learn. Because each individual subtask represents a relatively simpler function compared to the overall task of mapping raw sensors directly to driving decisions, each step can be learned with significantly less data than would be required to train a highly complex, end-to-end model.

Key points:

  • The pipeline breaks driving into three subtasks: detecting cars, detecting pedestrians, and path planning.
  • Each subtask represents a simpler function to learn.
  • Simpler functions can be learned with less training data.
  • A purely end-to-end approach requires more data because it attempts to learn a more complex function directly.

Rubric: The response must explain that the pipeline breaks the driving task into three subtasks. It must specify that each of these subtasks is a simpler function to learn. Finally, it must connect this simplicity to the fact that simpler functions require less training data than a purely end-to-end approach.

0

1

Updated 2026-05-26

Contributors are:

Who are from:

Tags

Machine Learning

Deep Learning

Supervised Learning

Dive into Deep Learning @ D2L

Data Science

Machine Learning Strategy

Related