Implement SVD from scratch (i.e. without using direct functions like numpy.linalg.svd) to visualize the famous 'king-man+woman=queen' equation.
The 300-dimensional word embeddings of 'king', 'queen', 'man' and 'woman' will be available in machineLearning/285 directory. Implement SVD from scratch, which should be applicable for all shapes of matrices. Use SVD to reduce the dimensionality of these 300-dimensional embeddings in order to visualize on a 2D plot.
You can use basic in-built functions related to matrix operations (like to obtain eigenvalues, eigenvectors or inverses). This exercise should help you in understanding more about matrix theory and basic concepts of word embeddings.
Details
Technical Specifications: Python
Type of issue: Single
Time Limit: 4 days
Issue requirements / progress
[ ] Python (knowledge about math-related modules like numpy and visualization modules like matplotlib)
Create a new folder called "285" inside machineLearning directory and add all code related to this issue in the machineLearning/285 folder
Note
Please claim the issue first by commenting here before starting to work on it. Please reference this issue properly in your PR. While submitting PRs, request review from @SakshatRao or @ameyanrd. Doubts or suggestions are welcome.
Description
Implement SVD from scratch (i.e. without using direct functions like numpy.linalg.svd) to visualize the famous 'king-man+woman=queen' equation.
The 300-dimensional word embeddings of 'king', 'queen', 'man' and 'woman' will be available in machineLearning/285 directory. Implement SVD from scratch, which should be applicable for all shapes of matrices. Use SVD to reduce the dimensionality of these 300-dimensional embeddings in order to visualize on a 2D plot.
You can use basic in-built functions related to matrix operations (like to obtain eigenvalues, eigenvectors or inverses). This exercise should help you in understanding more about matrix theory and basic concepts of word embeddings.
Details
Issue requirements / progress
Resources
Directory Structure
Create a new folder called "285" inside machineLearning directory and add all code related to this issue in the machineLearning/285 folder
Note
Please claim the issue first by commenting here before starting to work on it. Please reference this issue properly in your PR. While submitting PRs, request review from @SakshatRao or @ameyanrd. Doubts or suggestions are welcome.