实用深度学习(Practical Deep Learning)

Welcome!

Practical Deep Learning for Coders 2022 part 1, recorded at the University of Queensland, covers topics such as how to:

There are 9 lessons, and each lesson is around 90 minutes long. The course is based on our 5-star rated book, which is freely available online.

You don’t need any special hardware or software — we’ll show you how to use free resources for both building and deploying models. You don’t need any university math either — we’ll teach you the calculus and linear algebra you need during the course.

Real results

Our videos have been viewed over 6,000,000 times already! Take a look at the dozens of testimonials about our book and course by alumni, top academics, and industry experts.

By the end of the second lesson, you will have built and deployed your own deep learning model on data you collect. Many students post their course projects to our forum; you can view them here. For instance, if there’s an unknown dinosaur in your backyard, maybe you need this dinosaur classifier!

Alumni of our course have gone on to jobs at organizations like Google Brain, OpenAI, Adobe, Amazon, and Tesla, published research at top conferences such as NeurIPS, and created startups using skills they learned here. Petro Cuenca, lead developer of the widely-acclaimed Camera+ app, after completing the course went on to add deep learning features to his product, which was then featured by Apple for its “machine learning magic”.

Your teacher

I am the co-founder, along with Dr. Rachel Thomas, of fast.ai, the organization behind this course. At fast.ai we care a lot about teaching. In this course, I start by showing how to use a complete, working, very usable, state-of-the-art deep learning network to solve real-world problems, using simple, expressive tools. And then we gradually dig deeper and deeper into understanding how those tools are made, and how the tools that make those tools are made, and so on… We always teach through examples. We ensure that there is a context and a purpose that you can understand intuitively, rather than starting with algebraic symbol manipulation.

Is this course for me?

Previous fast.ai courses have been studied by hundreds of thousands of students, from all walks of life, from all parts of the world. Many students have told us about how they’ve become multiple gold medal winners of international machine learning competitions, received offers from top companies, and having research papers published. For instance, Isaac Dimitrovsky told us that he had “been playing around with ML for a couple of years without really grokking it… [then] went through the fast.ai part 1 course late last year, and it clicked for me”. He went on to achieve first place in the prestigious international RA2-DREAM Challenge competition! He developed a multistage deep learning method for scoring radiographic hand and foot joint damage in rheumatoid arthritis, taking advantage of the fastai library.

It doesn’t matter if you don’t come from a technical or a mathematical background (though it’s okay if you do too!); we wrote this course to make deep learning accessible to as many people as possible. The only prerequisite is that you know how to code (a year of experience is enough), preferably in Python, and that you have at least followed a high school math course.

Deep learning is a computer technique to extract and transform data–-with use cases ranging from human speech recognition to animal imagery classification–-by using multiple layers of neural networks. A lot of people assume that you need all kinds of hard-to-find stuff to get great results with deep learning, but as you’ll see in this course, those people are wrong. Here’s a few things you absolutely don’t need to do world-class deep learning:

The software you will be using

In this course, you’ll be using PyTorch, fastai, Hugging Face Transformers, and Gradio.

We’ve completed hundreds of machine learning projects using dozens of different packages, and many different programming languages. At fast.ai, we have written courses using most of the main deep learning and machine learning packages used today. We spent over a thousand hours testing PyTorch before deciding that we would use it for future courses, software development, and research. PyTorch is now the world’s fastest-growing deep learning library and is already used for most research papers at top conferences.

PyTorch works best as a low-level foundation library, providing the basic operations for higher-level functionality. The fastai library one of the most popular libraries for adding this higher-level functionality on top of PyTorch. In this course, as we go deeper and deeper into the foundations of deep learning, we will also go deeper and deeper into the layers of fastai.

Transformers is a popular library focused on natural language processing (NLP) using transformers models. In the course you’ll see how to create a cutting-edge transfomers model using this library to detect similar concepts in patent applications.

Why deep learning?

Deep learning has power, flexibility, and simplicity. That’s why we believe it should be applied across many disciplines. These include the social and physical sciences, the arts, medicine, finance, scientific research, and many more. Here’s a list of some of the thousands of tasks in different areas at which deep learning, or methods heavily using deep learning, is now the best in the world:

• Natural language processing (NLP) Answering questions; speech recognition; summarizing documents; classifying documents; finding names, dates, etc. in documents; searching for articles mentioning a concept
• Computer vision Satellite and drone imagery interpretation (e.g., for disaster resilience); face recognition; image captioning; reading traffic signs; locating pedestrians and vehicles in autonomous vehicles
• Medicine Finding anomalies in radiology images, including CT, MRI, and X-ray images; counting features in pathology slides; measuring features in ultrasounds; diagnosing diabetic retinopathy
• Biology Folding proteins; classifying proteins; many genomics tasks, such as tumor-normal sequencing and classifying clinically actionable genetic mutations; cell classification; analyzing protein/protein interactions
• Image generation Colorizing images; increasing image resolution; removing noise from images; converting images to art in the style of famous artists
• Recommendation systems Web search; product recommendations; home page layout
• Playing games Chess, Go, most Atari video games, and many real-time strategy games
• Robotics Handling objects that are challenging to locate (e.g., transparent, shiny, lacking texture) or hard to pick up
• Other applications Financial and logistical forecasting, text to speech, and much more…

What you will learn

After finishing this course you will know:

• How to train models that achieve state-of-the-art results in:
  • Computer vision, including image classification (e.g., classifying pet photos by breed)
  • Natural language processing (NLP), including document classification (e.g., movie review sentiment analysis) and phrase similarity
  • Tabular data with categorical data, continuous data, and mixed data
  • Collaborative filtering (e.g., movie recommendation)
• How to turn your models into web applications, and deploy them
• Why and how deep learning models work, and how to use that knowledge to improve the accuracy, speed, and reliability of your models
• The latest deep learning techniques that really matter in practice
• How to implement stochastic gradient descent and a complete training loop from scratch

Here are some of the techniques covered (don’t worry if none of these words mean anything to you yet–you’ll learn them all soon):

• Random forests and gradient boosting
• Affine functions and nonlinearities
• Parameters and activations
• Transfer learning
• Stochastic gradient descent (SGD)
• Data augmentation
• Weight decay
• Image classification
• Entity and word embeddings
• And much more

How do I get started?

To watch the videos, click on the Lessons section in the navigation sidebar. The videos are all captioned; while watching the video click the “CC” button to turn them on and off. To get a sense of what’s covered in a lesson, you might want to skim through some lesson notes taken by one of our students (thanks Daniel!). Here’s his lesson 7 notes and lesson 8 notes. You can also access all the videos through this YouTube playlist.

Each video is designed to go with various chapters from the book. The entirety of every chapter of the book is available as an interactive Jupyter Notebook. Jupyter Notebook is the most popular tool for doing data science in Python, for good reason. It is powerful, flexible, and easy to use. We think you will love it! Since the most important thing for learning deep learning is writing code and experimenting, it’s important that you have a great platform for experimenting with code.

We’ll mainly use Kaggle Notebooks and Paperspace Gradient because we’ve found they work really well for this course, and have good free options. We also will do some parts of the course on your own laptop. (If you don’t have a Paperspace account yet, sign up with this link to get $10 credit – and we get a credit too.)

We strongly suggest not using your own computer for training models in this course, unless you’re very experienced with Linux system adminstration and handling GPU drivers, CUDA, and so forth.

If you need help, there’s a wonderful online community ready to help you at forums.fast.ai. Before asking a question on the forums, search carefully to see if your question has been answered before.