Case context: A machine learning team is working on two different image-based tasks: Task A is simply identifying if an image is dark or light, which their current model solves effectively using a very shallow neural network. Task B involves classifying 1,000 distinct object categories from complex backgrounds, which requires a highly multi-layered neural network to achieve acceptable accuracy.

Question: Based on the informal definitions of task difficulty in machine learning, how should the team classify Task A and Task B, and what reasoning supports this classification?

Sample answer: Based on the informal definitions, Task A is an "easy" task because it requires fewer computation steps and is handled by a shallow neural network. Task B is a "hard" task because it requires many computation steps and necessitates a deeper neural network.

Key points:

• Task A is informally classified as an easy task.
• Easy tasks can be carried out with fewer computation steps (shallow networks).
• Task B is informally classified as a hard task.
• Hard tasks require more computation steps (deeper networks).

Rubric: The answer must identify Task A as easy and Task B as hard, explicitly tying these classifications to the shallow and deep neural network architectures, respectively.