ML//neural network//perceptron

The original single-neuron classifier (Rosenblatt, 1958). Weights each input, sums, adds a bias, and thresholds.

The original single-neuron classifier (Rosenblatt, 1958). Weights each input, sums, adds a bias, and thresholds.

Computes z = w·x + b, then fires 1 if z clears a threshold, else 0. Geometrically it draws a single hyperplane: one side "yes", the other "no".

Learns by nudging weights toward less-wrong after each mistake, the seed of all gradient descent training.

Hard ceiling: one hyperplane only separates linearly separable data (it cannot learn XOR). The fix is stacking neurons into a neural network.