This section presents an example of implementing kernel methods in machine learning, demonstrated through a classification task using the Fashion MNIST dataset (10 classes).
The project features the development and application of an SVM algorithm with kernel trick (using kernel approximation through the method of RFF — Random Fourier Features).
The work also includes a comparison of different classification algorithms:
- Kernel and Linear SVM
- Logistic Regression
- Gradient Boosting with CatBoost implementation
Examples are provided showing all algorithms operating with and without dimensionality reduction.