Interactive MNIST Forward Propagation
1. Input (28 × 28 = 784)
Draw above or hover over a pixel.
X = (x1, x2, …, x784)
2. Network Architecture (Click a Hidden Neuron)
Input Layer (784) → Hidden Layers (ReLU) → Output Layer (10)
3. Output (Softmax 0–9)
Multi-class probabilities sum to 1.00.
Selected Hidden Neuron (H1, Neuron 1): Calculating weighted sum z = Σ wixi + b and applying ReLU activation σ(z) = max(0, z).