Neural Networks

• 

  Architecture for Neural Networks

This project was inspired by the desire to understand how neural networks work internally instead of only using machine learning libraries. The project builds a neural network completely from scratch using Python to solve the XOR problem. It takes two inputs and learns to predict the correct output through training. All core parts such as weights, biases, activation function, forward propagation, loss calculation, and backpropagation were implemented manually without using libraries like TensorFlow or PyTorch. One of the main challenges was understanding and correctly implementing backpropagation and tuning the learning rate. Despite this, the network was successfully trained to predict the XOR outputs correctly. Through this project, we gained a strong understanding of how neural networks learn step by step and improved our problem-solving skills. In the future, this project can be extended by adding more layers, trying different activation functions, and applying the network to more complex problems.

Built With

• numpy
• python
• vs-code