Scribd
Upload
3 views2 pages

Semester Project

The CS-844 Semester Project is due on May 1st, 2026, requiring groups of up to three students or individual submissions. Students must conduct a literature review on two recent research papers related to Generative Deep Models and submit their choices by April 15th, 2026. The evaluation will consist of a viva (50%), report (30%), and implementation (20%) based on the selected papers.

Uploaded by

sohail
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
3 views2 pages

Semester Project

The CS-844 Semester Project is due on May 1st, 2026, requiring groups of up to three students or individual submissions. Students must conduct a literature review on two recent research papers related to Generative Deep Models and submit their choices by April 15th, 2026. The evaluation will consist of a viva (50%), report (30%), and implementation (20%) based on the selected papers.

Uploaded by

sohail
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

CS-844 Semester Project

Due Date May 1st, 2026


Deliverable: A pdf on LMS. A demonstration of the code during viva.
Groups: Groups of at most 3 people need to be formed. You can do the
project individually as well but there is no extra grade for that. Please
submit your groups using this link:
[Link]
Evaluation: A viva will be conducted after the due date. Grading will be
done on an individual basis.
Viva – 50%
Report – 30%
Implementation – 20%
Description:
This is primarily an open-ended research project. The main purpose is for
you to spend some time doing literature review and gain a deeper
understanding of specific Generative Deep Models. You can use
components of your ongoing research if applicable.
Literature Review:
You have to select TWO research papers with the following constraints:
1. The papers should be relatively recent. No more than 6 years old.
2. The papers should be published in well known machine learning
venues such as NeurIPS, ICML, ICLR, CVPR, JMLR, TPAMI etc. or
have a significant number (>100) citations.
3. The papers should be related to Generative Deep Models.
Please submit your group’s choice of the two papers by April 15th, 2026
using the following link: [Link]
Report:
Once you have selected the two papers you have to write a 3 to 4 page
review of each paper covering the following:
1. A short but comprehensive summary of the paper.
2. What kind of Generative Deep Model is used / proposed in the paper.
3. What makes the proposed method better than the previous
available methods. Especially what makes it better than the basic
versions of the same type of models.
4. What are the downsides of the proposed method.
5. What improvements can be made (you can look at papers published
after the selected paper for help).
Implementation:
After that, you will have to choose one of these papers and implement
them yourself. Be mindful of resource constraints that you might run into.
E.g. if you are working with a method that generates images, you can
consider implementing a simpler/smaller version of it and train it on
simpler datasets like MNIST / CIFAR10 etc.

You might also like