This repo contains the code for the implementations of various classifiers
on the popular MNIST dataset. source_code_p1.ipynb
contains the code for
implementing a kernel perceptron and running multiple experiments on the dataset.
source_code_p2.ipynb
contains the manual implementations for 4 basic algorithms:
the perceptron, the winnow, least squares regression and K-nearest Neighbors. Lastly,
p1_svm_manual_implementation.ipynb
contains a manual implementation of the support vector
classifier, which is connected to the task in source_code_p1.ipynb
.
The results for all my models are accessible in the folders part_1_results/
and part_2_results/
.
For a detailed discussion of my methods and results, checkout my report
in report.pdf
.
- NumPy
- CuPy
- SciPy
- MatPlotLib
- Seaborn
- Pandas
- CVXOPT
- itertools, os, time