Week 8 - Deep Learning: Advanced Concepts

Question 1

What is Saliency in the context of Computer Vision?

Answer 1

Saliency networks try and predict this map of interesting areas in the image, which aims to simulate human attention.

Question 2

What are the definitions for local and global context?

Answer 2

Local - Detailed neighbourhood attention
Global - Where the object sits in the scene

Question 3

How does an unsupervised GAN work?

Answer 3

The aim is for the generator to fool the discriminator network, so that it can’t tell the difference between real and fake.

Question 4

What are some problems with GANs?

Answer 4

They are tough to train:
- If the discriminator behaves badly, then the generator does not have accurate feedback and the loss functions cannot represent the reality
- If the discriminator does a great job, then the gradient of the loss function drops down close to 0, and the learning becomes super slow or even jammed.

Mode Collapse:
- During the training, the generator may collapse to a setting where it produces similar outputs with low variety