$1.$ VAEs and Diffusion Models

I. Discuss the difference between them and their

potential uses in target applications.

II$.$ Why is one better than the other for a given

application.

$3.$ What is the purpose of Positional encoding in the context of a

transformer?

$4.$ Describe a Multi$-$modal Transformer used within the context of

medical diagnosis or Emotion Recognition?

$5.$ Consider a dataset of brain image segments $($e$.$g$.,$ CAT scans$)$;

discuss how a combination of generative models can be used for anomaly

detection.

$6.$ Consider mapping Brain signals to its corresponding image that

is visualised;

I. Discuss how Diffusion models can be used in such a

use$-$case.

II$.$ Relate to the challenges and how they can be overcome

using advanced composition of deep learning architecture.

$3.$ Describe the difference between Supervised, Unsupervised and

Self Supervised learning in the context of a deep learning model such as a

VAE

$4.$ Describe a use$-$case along with the dataset for a Graph NN

$5.$ Could you explain the concept of data fusion and the

fundamental necessity for implementing a cross$-$attention mechanism

$6.$ Discuss the ways in which Large Language Models $($LLMs$)$ might

augment or enhance your model