Applied Deep Learning
Apply deep learning to real problems: fast.ai's practical course, computer vision with OpenCV, perception for self-driving cars and AlphaZero from scratch.
Start learning Open course bookAbout this course
Knowing how neural networks work is the start; applying them well is a separate skill. Real projects involve choosing pre-trained models, cleaning data, processing images, combining several models into a system and, in some cases, learning without any labelled data at all.
This course follows four applied freeCodeCamp courses. Jeremy Howard and Sylvain Gugger's Practical Deep Learning for Coders teaches the fast.ai top-down approach, from first models to production, ethics and recommendation systems. Jason Dsouza's OpenCV course covers classic image processing and detection. Sakshay Mahna's course surveys the perception tasks of a self-driving car, and Robert Förster builds AlphaZero, which learns games by playing against itself. Each lesson comes with an original book chapter of worked examples.
What you will learn
- Fine-tune pre-trained networks and take a model into production responsibly
- Process images with OpenCV: colour spaces, blurring, thresholds, edges and contours
- Explain the perception tasks of a self-driving car, from segmentation to tracking
- Compute detection and tracking quantities such as intersection over union by hand
- Describe how Monte Carlo tree search and self-play train AlphaZero
Who this course is for
Programmers who know the basics of deep learning and Python and want to apply them. Earlier courses on this site, especially Deep Learning with PyTorch, are good preparation.