Scribd
Upload
32 views

CSL7620_A2

The document outlines the requirements for Assignment 2 of the Machine Learning course CSL7620, due on April 13, 2025. It includes instructions for report formatting, submission guidelines, and penalties for non-compliance, along with specific implementation questions involving datasets and machine learning techniques. The assignment is divided into two parts, focusing on practical implementation and a project requiring an end-to-end machine learning pipeline.

Uploaded by

muskaanbhayana9
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
32 views

CSL7620_A2

The document outlines the requirements for Assignment 2 of the Machine Learning course CSL7620, due on April 13, 2025. It includes instructions for report formatting, submission guidelines, and penalties for non-compliance, along with specific implementation questions involving datasets and machine learning techniques. The assignment is divided into two parts, focusing on practical implementation and a project requiring an end-to-end machine learning pipeline.

Uploaded by

muskaanbhayana9
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/ 2

Machine Learning - CSL7620

Assignment 2 (50 marks): Weightage 15%


6th March 2025
Date of Submission: 13th April 2025; 23:59

Instructions (Non-conformance will result in penalties):


●​ A typed report must be written for the assignment (no handwritten reports will be
accepted). Please ensure the report is neat, properly formatted, and contains all relevant
information to the tasks in the assignment.
●​ Grading of the assignment shall be done based on both the codes and the report. The
report shall reflect your understanding regarding the concepts, methods that you have
used, and the analysis of the results. Please keep the content of your reports crisp and
precise. Max. 12 page report.
●​ Submission should contain the following files - one .ipynb file to LMS and report (.pdf).
to Turnitin (link will be shared later). Please name each of these as roll_number(s).ipynb
and roll_number(s).pdf.
●​ Within the code notebook, please make proper sections for each question as discussed in
the demonstration session.
●​ Any deviation from the above-mentioned submission format or deadline breach will
result in a penalty of marks.
●​ Even though intellectual discussion is encouraged, any plagiarism within the report is
completely unacceptable and will result in the nullification of marks and a possible action
for academic misconduct (according to the relevant institute regulations).
●​ You can do this in groups of either 2 or 3. Mention the contributions of all members.

Part 1: Implementation Questions


Q1: Download the following dataset diabetes (1).csv - Google Drive. ( 10 Marks )
a)​ Find the optimum number of principal components for the features in the
above-mentioned data
b)​ Use any two regression models of your choice and find the prediction accuracy and error
between the reduced data (with an optimum number of principal components) and the
complete data.

Q2: We will use the fashion-MNIST dataset for this question (you can download it from any
other source also including libraries). Flatten and preprocess the data (if required) before starting
the tasks. It will become a 784-dimensional data with 10 classes, more details are available in the
link. ( 20 Marks )
a)​ Train the k-means model on f-MNIST data with k = 10 and 10 random 784-dimensional
points (in input range) as initializations. Report the number of points in each cluster.
b)​ Visualize the cluster centers of each cluster as 2D images of all clusters.
c)​ Visualize 10 images corresponding to each cluster.
d)​ Train another k-means model with 10 images from each class as initializations , report the
number of points in each cluster, and visualize the cluster centers.
e)​ Visualize 10 images corresponding to each cluster.
f)​ Evaluate Clusters of part a and part d with Sum of Squared Error (SSE) method. Report
the scores and comment on which case is a better clustering.

Q3: Implementation of Neural Networks from Scratch Using NumPy and Comparison with
Sklearn (20 marks)
a)​ Load and preprocess the MNIST Digits Dataset. (3 marks)
b)​ Implement a neural network with one input layer, one hidden layer, and one output layer
using NumPy. (5 marks)
c)​ Train the neural network with various hyperparameters (e.g., learning rate, number of
hidden nodes). (3 marks)
d)​ Evaluate the performance of the neural network on the testing set. (2 marks)
e)​ Implement the same neural network using sklearn and compare the results with the
NumPy implementation. (4 marks)
f)​ Plot the training and validation loss/accuracy curves (for both experiments). (3 marks)

Part 2: Project Part

Q4: Select a project from the list provided in the document and implement a complete
end-to-end machine learning pipeline for the same. Also, prepare a demo using gradio/streamlit
for evaluation. For gradio sample app ref:
​https://colab.research.google.com/github/kirenz/lab-huggingface/blob/main/code/gradio.ipynb
(50 marks)
List of projects:
https://docs.google.com/document/d/1HYz5TA1QBhhutvWdKIKvTr4v6SzebjrKxZbklh8zsZo/e
dit?usp=sharing

Note: The report should contain detailed explanations and analysis for your observations. Just
reiterating the code will not fetch you any marks.

You might also like