KEMBAR78
AML2023 Assignment 2 GAN | PDF
Scribd
Upload
11 views1 page

AML2023 Assignment 2 GAN

The assignment requires the construction of a variational autoencoder for the Fashion MNIST dataset, a GAN for the MNIST dataset, and a Conditional GAN for MNIST, with specific image generation tasks outlined for each model. Students must submit a Jupyter notebook with code and outputs, along with a one-page write-up documenting their work. Collaboration in groups of 2 or 3 is allowed, but only one submission per group is required, and timely submission confirmation with TAs is necessary.

Uploaded by

abhishek
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
11 views1 page

AML2023 Assignment 2 GAN

The assignment requires the construction of a variational autoencoder for the Fashion MNIST dataset, a GAN for the MNIST dataset, and a Conditional GAN for MNIST, with specific image generation tasks outlined for each model. Students must submit a Jupyter notebook with code and outputs, along with a one-page write-up documenting their work. Collaboration in groups of 2 or 3 is allowed, but only one submission per group is required, and timely submission confirmation with TAs is necessary.

Uploaded by

abhishek
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 1

AML 2023: Assignment 2

CMI

Due Date: Nov 10, 2023

1. The Fashion MNIST dataset consists of 28x28 grayscale images from 10 categories of
images of clothes, shoes etc. Construct a variational autoencoder for them.
https://www.kaggle.com/datasets/zalando-research/fashionmnist
a. Show 5 examples of images produced by random samples from the latent space
b. Show 5 examples of transitions between two images in the dataset, as we saw in
class.

2. Create a GAN for MNIST dataset which has 28x28 grayscale images of 10 digits.
https://www.kaggle.com/datasets/hojjatk/mnist-dataset
a. Show 10 images generated from random codewords.
b. Show 5 examples of the following: Pick two random codewords, a and b, and
pick 10 uniformly spaced points between a and b on the line from a to b. Show
the images generated from these 12 points

3. Create a Conditional GAN for MNIST.


- Recall that, in a conditional GAN, we give the generator a random codeword from
the latent space, as well as a label indicating which class of image we want, and
the generator tries to create an image of this class.
- The Discriminator is given pairs of (Image, Label) to classify as real or fake
- For each class, show 4 examples of generated images from random codewords

Instructions:
- You have to submit a jupyter notebook (ipynb) with all your code and outputs of the code
- You should also submit a 1 page writeup documenting what you have done
- If you don’t include the outputs you will get partial credit
- You can work in groups of 2 or 3
- Only one member of the group should submit the assignment
- Please mention the names and roll-numbers of all group members
- You are free to build upon examples shown in class
- Please confirm with the TAs that your submission has been received on time.
- No requests for re-submitting the assignment later because of various reasons will be
entertained.

You might also like