In this project, I implemented, tested, and evaluated a system capable of classifying MNIST images using thoroughly unit-tested data structures and algorithms in C++.
I implemented templated List, Vector, Tree, and Table data structures to store and process information. I also wrote directed and random unit-tests for each of the subsystems to deploy a thoroughly tested handwriting recognition system. I then profiled the performance of the system using flamegraphs and applied optimizations to speed up execution 16 fold.
A key theme of this project is that no inbuilt libraries were used. This was done to build a better understanding of how data structures and algorithms work under the hood in C++.
The zip file includes 4 reports from 4 sections of the project, as well as a testing strategy document that outlines my testing strategy throughout this project.