Bihar STET Computer Science – Artificial Intelligence Set 1

K-Means is an unsupervised learning algorithm that partitions unlabelled data points into K distinct clusters based on feature similarity.

Gradient Descent updates parameters $\theta := \theta - \alpha * grad(J(\theta))$ to minimize the loss function J.

Backpropagation efficiently applies the partial derivative Chain Rule from output back to input layers to compute loss gradients for weight updates.

A Perceptron is the simplest form of a neural network unit, invented by Frank Rosenblatt, taking weighted inputs and applying a threshold activation function.

ReLU outputs x if x > 0, and 0 otherwise, defined as f(x) = max(0, x), preventing vanishing gradient issues in deep networks.