Machine Learning with Deep Learning in Urdu (recluze)

A lecture-style Urdu course of 17 long sessions, about 20 hours, that goes from classical machine learning into deep learning. It opens with demos, tools and notation, then scikit-learn, linear regression and optimisation, gradient descent with contour plots, a practical PyTorch regression, classification, multiclass problems and regularisation. The second half covers neural networks and back-propagation, CNNs, RNNs, LSTMs, model selection, a probability reboot for ML, unsupervised learning with neural networks and GANs. Suited to university students who want a connected path from regression to modern deep learning in one series.