I made this project because I want to understand how Neural Network actually works under the hood and how it related to linear algebra.
- CMAKE
- C/C++ Compiler
- Build tool
git clone <repo-url>
on linux
cd Neural-Network-From-Scratch
mkdir build
cd build
cmake .. -DCMAKE_BUILD_TYPE=Release
make
on windows
cd Neural-Network-From-Scratch
mkdir build
cd build
cmake -G "MinGW Makefiles" -DCMAKE_BUILD_TYPE=Release ..
mingw32-make
cd ../output
./Neural_Network
Note: Usually, it takes round 45 minutes to complete.
MNIST datasets
Gradient Decent blog
Learn to build a Neural Network from scratch blog
Neural Network explanation playlists
Softmax blog explanation
LeakyRelu blog explanation
Cross entropy loss blog explanation
Note: In this project, I used Gemini to help me learn about other activation functions suitable for MNIST datasets, and I also used Gemini to help me learn how to write proper C++.
I also want to thank you for all the high-quality blogs and videos created for everyone to learn from. Finally if there are any mistakes please let me know and I will fix them thank you for your attention.