Case Study

Allocating resources for a startup's first product dev and test sets.

Case context: A startup is building a novel computer vision model to detect defects in manufacturing. They have limited funding and must decide whether to spend 10% or 50% of their initial budget on meticulously curating their dev and test sets.

Question: Based on the principle that dev and test set investment requires judgment, what factors should the startup's leadership consider when deciding how much of their budget to allocate to this task?

Sample answer: The leadership must use their judgment to balance the need for a highly reliable evaluation metric against their limited budget. They should consider how critical the defect detection is (e.g., the real-world cost of a false negative), how difficult the data is to acquire, and whether a smaller, less expensive dev/test set could still adequately reflect the future data they expect to see in production.

Key points:

  • Balance budget constraints with the need for accurate evaluation.
  • Consider the real-world criticality of the application.
  • Ensure the sets reflect future data despite resource limitations.

Rubric: The response should identify the need to balance resources against the necessity of accurate evaluation, considering the specific context of the startup's constraints and application criticality.

0

1

Updated 2026-05-27

Contributors are:

Who are from:

Tags

Machine Learning

Deep Learning

Machine Learning Strategy

Supervised Learning

Dive into Deep Learning @ D2L

Data Science

Machine Learning Yearning @ DeepLearning.AI