Gradient Descent & Momentum
How models learn: step downhill on a loss surface; learning rate, momentum, the Rosenbrock valley.
Zahri-Edu › Math-AI
The mathematics behind artificial intelligence: optimisation, regression, neural networks and learning.
How models learn: step downhill on a loss surface; learning rate, momentum, the Rosenbrock valley.
Fit a line to your own points; MSE and R² from the normal equations.
A linear classifier learning from mistakes, and why XOR needs more layers.
Unsupervised learning: assign, update, repeat.
Back-propagation training a hidden layer on circles, XOR and spirals.
Sigmoid, tanh, ReLU and the softmax temperature used by language models.
Lecture on the mathematical foundations of artificial intelligence (in Arabic).
Energy functions, probability and learning.
AI and the human anatomy through symmetry.
Swarms and multi-agent systems.
Linear algebra, calculus, probability and optimisation for machine learning.