Backpropagation (THE HOW)

---

\[ \begin{array}{l} \text{Backpropagation Algorithm} \\ \text{epochs} = 5 \to (5) \\ \text{for } i \text{ in range(epochs):} \\ \quad \text{for } j \text{ in range}(\mathbf{X}.\text{shape}[0]): \\ \quad \quad \to \text{ Select 1 row (random)} \\ \quad \quad \to \text{ Predict (using forward prop)} \\ \quad \quad \to \text{ Calculate loss (using loss function } \to \text{ mse)} \\ \quad \quad \to \text{ Update weights and bias using GD} \\ \quad \quad \quad \mathbf{W}_{\text{new}} = \mathbf{W}_{\text{old}} - \eta \frac{\partial L}{\partial \mathbf{W}} \\ \quad \to \text{ Calculate avg loss for the epoch} \end{array} \]