NumpyNeuralNetwork: Building Neural Networks from Scratch with NumPy

Numpy Neural Network, where I dive deep into the world of neural networks using NumPy! In this repository, you'll find my Code for creating neural networks from the ground up, without relying on external libraries.

Neural Network Formula (FeedForward)

$$z = \sigma \sum_{i=1}^{n}\left ( \vec{x_i}\vec{w_i} \right ) + b$$

• $\sigma$ : Activation Function
• $x_i$ : Input Data
• $w_i$ : Weight
• $b$ : Bias

Lost / Cost Function (Mean Squared Error)

$$MSE = \frac{1}{n}\sum_{i = 1}^{n}(Y_i - \hat{Y_i})^2$$

• $Y_i$ : real value
• $\hat{Y_i}$ : predicted value

Update weight (gradient descent)

$$W' = W + \alpha \frac{dL}{dW}$$

$$b' = b + \alpha \frac{dL}{db}$$

• $\alpha$ : learning rate
• $L$ : Lost function