Deep learning is the branch of machine learning behind modern computer vision, speech and language models. This module introduces neural networks, how they learn through gradient descent and backpropagation, and the architectures built on top of them, including convolutional networks for images and residual networks.
The lessons are hands-on with PyTorch, the framework most deep learning research is written in: tensors, building and training models, using GPUs, regularization and data augmentation, and generative adversarial networks (GANs).
These courses are long, so treat them as a reference: watch the parts you need, then come back for the rest. When you're done, continue on to Transformers & NLP.
How the math connects to neural network architecture, activation functions, and backpropagation.
PyTorch basics and linear regression, image classification with logistic regression, training deep neural networks on a GPU, convolutional neural networks, residual networks, data augmentation and regularization, and generative adversarial networks (GANs).
A code-first course: tensors and PyTorch fundamentals, the PyTorch workflow, neural network classification, computer vision, and custom datasets, building toward a food image classification project.