Scribd
Upload
6 views3 pages

MML Assignment

fff

Uploaded by

Akshaya
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
6 views3 pages

MML Assignment

fff

Uploaded by

Akshaya
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/ 3

Assignment 1

Due date: December 2, 2024

1. Building a Simple Linear Regression Model Using Projections


Problem Description: Use orthogonal projection to derive the least-squares solution
for a simple linear regression problem.
– Dataset: Predict house prices based on a single feature (e.g., size).
– Fit a line to the data by projecting data points onto the column space of the feature
matrix.
Questions:
– Why is the orthogonal complement important in the derivation of the least-squares
solution?
– Compute the norm of the residuals (error vector). How does the norm change with
additional features?
– Visualize the projection and explain its geometric interpretation.
2. Image Compression Using Singular Value Decomposition (SVD)
Problem Description: Compress a grayscale image using SVD.
– Decompose the image matrix into U ΣV T .
– Reconstruct the image using only the top k singular values.
Questions:
– How does the reconstruction quality vary with k?
– Compare the storage size of the original matrix versus the decomposed represen-
tation.
– What do the left and right singular vectors represent in the context of the image?
3. Eigenvalues and Eigenvectors in Graph Analysis
Problem Description: Analyze a small social network graph using its adjacency
matrix.
– Compute the eigenvalues and eigenvectors of the matrix.
– Identify the importance of nodes based on eigenvector centrality.
Questions:
– What does the largest eigenvalue signify in the context of the graph?
– How can eigenvectors help in clustering nodes?
– Interpret the eigenvector centrality of a node in a simple social network.
4. Dimensionality Reduction Using PCA
Problem Description: Apply PCA to a high-dimensional dataset (e.g., MNIST or
any numerical dataset with multiple features).
– Center the dataset.
– Compute the covariance matrix and its eigenvalues/eigenvectors.
– Project the dataset onto the first two principal components.
Questions:
– What proportion of variance is retained by the first two principal components?
– How does dimensionality reduction affect clustering or classification tasks?
– Visualize the dataset in the reduced dimension and interpret the results.
5. Solving Systems of Equations Using Cholesky Decomposition
Problem Description: Solve a system of linear equations Ax = b, where A is sym-
metric positive definite, using Cholesky decomposition.
– Implement the decomposition to factorize A = LLT .
– Solve for x in two steps: Ly = b, then LT x = y.
Questions:
– Why is Cholesky decomposition more efficient than other methods for solving such
systems?
– What are the computational advantages of this approach for large matrices?
– Verify the solution using standard methods like Gaussian elimination.

Assessment Rubric on a 10-Point Scale

The following rubric is used to evaluate each question based on a 10-point scale, with
categories listed below.

1. Mathematical Accuracy (3 points)

– 3 points: The solution demonstrates a thorough understanding of linear algebra con-


cepts, with correct application and detailed reasoning. All calculations are accurate.
– 2 points: The solution is mostly correct but contains minor errors or incomplete
reasoning in a few areas. The majority of the calculations are accurate.
– 1 point: The solution has significant errors in applying linear algebra concepts or
makes major calculation mistakes.
– 0 points: The solution is fundamentally incorrect, with little to no understanding of
linear algebra concepts.

2. Code Implementation (2 points)

– 2 points: Code is well-organized, efficient, and follows best practices for readability.
The implementation is correct and produces the expected output.
– 1 point: Code is functional but lacks efficiency or has minor readability issues. There
may be small mistakes in implementation.
– 0 points: Code is inefficient, disorganized, or produces incorrect results.
3. Analysis and Interpretation (2 points)

– 2 points: Provides a thorough analysis and interpretation of results. The discussion


is insightful, highlighting important patterns or conclusions drawn from the results.
– 1 point: Provides a basic analysis with some correct interpretations, but may overlook
key aspects or miss deeper insights.
– 0 points: Little to no analysis provided, or the analysis is inaccurate or incomplete.

4. Visualization (1 point)

– 1 point: High-quality visualizations that are clear, informative, and enhance the
understanding of the solution. Graphs and plots are appropriately labeled and easy
to interpret.
– 0 points: No visualizations, or poor-quality visuals that do not contribute to under-
standing the solution.

5. Documentation (2 points)

– 2 points: Clear, concise, and well-structured documentation. The problem, approach,


and results are explained thoroughly, with good use of examples and mathematical
notations.
– 1 point: Documentation is present but lacks clarity, depth, or structure in explaining
the process or results.
– 0 points: Little to no documentation, or the documentation is unclear and unhelpful.

Total Score: 10 points

You might also like