A collection of ML projects completed through my Masters degree coursework. Many projects involved writing ML algorithms from scratch.
| Project | Description |
|---|---|
| K-Means: Image Compression | This project builds the k-means algorithm from scratch, and uses it to compress colored images. In a colored image, each pixel has a size of 3 bytes (RGB), where each color is represented by a value from 0 to 255. Using the k-means clustering algorithm, we can take advantage of the visual perception of the human eye by clustering the most similar pixels (colors) in an image as data points, and visualize only the most prominent colors while maintaining the structure of the image and decreasing the size of the image file.  |
| PCA: Food Consumption in European Countries | Implementing Principal Component Analysis (PCA) from scratch on data consisting of 16 European countries and their consumption for 20 food items, such as tea, jam, coffee, etc.  |
| ISOMAP: Non-linear Dimensionality Reduction | ISOMAP is a non-linear dimension reduction method to reduce the size and complexity of a dataset, projecting it onto a new plane. This method is useful for datasets with non-linear structure, where PCA and MDS will not be appropriate |
| PCA: Facial Recognition with Yale Eigenfaces | Using facial images from The Yale Face Database B, where there are 64 images under different lighting conditions per each of 10 distinct subjects, 640 face images in total. |