Limits of Deep Learning

Lesson goals

• Data and compute demands
• Interpretability limits
• Distribution shift

Core idea

Cost, interpretability, brittleness, and the need for evaluation.

1. Data and compute demands: understand it in context, identify inputs and outputs, and define how the result will be verified.

2. Interpretability limits: understand it in context, identify inputs and outputs, and define how the result will be verified.

3. Distribution shift: understand it in context, identify inputs and outputs, and define how the result will be verified.

Practical example

Consider a real application of “Limits of Deep Learning.” Start small, identify available data, and compare the system with a clear baseline before expanding it.

Check your understanding

1. Explain “Limits of Deep Learning” in your own words.
2. Give one appropriate and one misleading application.
3. What risk or limitation should be monitored?