Classical machine learning works well on tables of features, but images, text, and speech need something more powerful. Deep learning uses layered neural networks that learn their own features directly from raw data, and it powers most of today's breakthroughs in AI.
You'll begin with how neural networks are built and how they learn, then see how convolutional neural networks (CNNs) handle images. From there you'll move into natural language processing (NLP): representing words with embeddings, and understanding transformers and the attention mechanism behind models like GPT.
Once you understand how these models work, the last step is getting them out of a notebook and into the real world in Deployment & MLOps, the next module in this course.
A visual introduction to neural networks through handwritten digit recognition: what neurons and layers are, and the math underneath.
How convolutional neural networks use filters to find patterns in images and classify them, explained step by step.
How words are turned into numbers that capture meaning, the starting point for most modern NLP, with a walkthrough of Word2Vec.
A visual introduction to transformers, the architecture behind GPT and modern language models, focusing on overall structure and word embeddings.
Demystifies attention, the key mechanism inside transformers and large language models, with clear animations.