考前速记

A* Algorithm

$f(n) = g(n) + h(n)$

	A*	Dijkstra
Heuristic	Yes (h(n))	No
Finds	Shortest path to goal	Shortest to all nodes
Guarantee	Only if admissible	Always

ML Types — Quick Check

Training data labelled?     → Supervised
Training data unlabelled?   → Unsupervised
Agent + reward/penalty?     → Reinforcement
Multiple hidden layers?     → Deep Learning

ANN Vocabulary

Term	Meaning
Weight	Strength of connection between neurons
Bias	Offset to shift activation
Activation fn	Introduces non-linearity (Sigmoid, ReLU)
Forward prop	Input → Hidden → Output
Back prop	Error propagated backwards to update weights
Learning rate ($\alpha$)	Controls step size in gradient descent

Command Words — Quick Guide

Word	Do
Calculate	Show all steps, fill table
Describe	List key features (2-3 points)
Explain	How/why it works (mechanism)
Identify	Name the type + brief reason
Justify	Evidence for your answer