Short Answer

Explain how human-designed features help when the training set is small.

Question: A team is building a voice-command system, but it only has a small labeled dataset. How can carefully designed input features, such as pitch summaries and boundary markers, work together with the learning algorithm?

Sample answer: Human-designed features add prior knowledge to the input representation. They give the learning algorithm a better starting point by packaging useful structure into the data, so the model does not need to discover everything from scratch. That makes learning easier and can improve performance when labeled examples are limited.

Key points:

  • Human-designed features add prior knowledge to the learned system
  • They simplify the learning task
  • They are especially helpful when labeled data is scarce
  • They let the model succeed with fewer examples

Rubric: The response must say that hand-designed knowledge adds to what the algorithm learns from data and that this is helpful when data is limited.

0

1

Updated 2026-08-12

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