This “Field Report” is a bit difference from all the other reports I’ve done for insideAINews.com because it is more of a “virtual” report that chronicles my experiences going through the content of an exciting new learning resource designed to get budding AI technologists jump started into the field of Deep Learning. Renowned MOOC platform Coursera just launched a new Deep Learning Specialization series consisting of 5 courses. I enrolled in the inaugural session and I’m now midway through the specialization. My purpose is to see what AI luminary Andrew Ng came up with to pull more people into the field, get a state-of-the-art glance at the field, and most importantly for my readers – assess the quality of this new learning resource. So far, I’ve been delighted to see a fresh, 2017 perspective on this exciting technology, and I intend to ride this specialization all the way to the end. Ng started a new organization called deeplearning.ai to produce the course content that uses Coursera as the learning platform. I have a lot of experience with Coursera having beta tested a cool Data Science specialization series from Johns Hopkins University a couple of years ago.
5 Course Series
All the courses are taught by Dr. Andrew Ng, Co-founder of Coursera, Adjunct Professor at Stanford University, and formerly head of Baidu AI Group/Google Brain. Here is a list of the classes in the program:
- Neural Networks and Deep Learning (4 weeks)
- Improving Deep Learning Networks: Hyperparameter tuning, Regularization and Optimization (3 weeks)
- Structuring Machine Learning Projects (2 weeks)
- Convolutional Neural Networks (TBD)
- Sequence Models (TBD)
If you want to break into AI, this specialization will help you do so. Deep learning is one of the most highly sought after skills in tech. In 5 courses, you will learn the foundations of deep learning, understand how to build neural networks, and learn how to lead successful machine learning projects. You will learn about Convolutional networks, RNNs, LSTM, Adam, Dropout, BatchNorm, Xavier/He initialization, and more. You will work on case studies from healthcare, autonomous driving, sign language reading, music generation, and natural language processing. You will master not only the theory, but also see how it is applied in industry. You will practice all these ideas in the Python language and the TensorFlow platform.
Other Features
Each class consists of a number (usually around 10) short video lectures or “screen casts.” Each week also has a 10 question quiz to help you assess your understanding of the material. Finally, you’ll have one or more programming assignments using pre-prepared Jupyter notebooks consisting of instructions, back ground theory and Python code along with expected output so you can check your work. The code consists of many Python functions that include clearly marked sections that you must fill in yourself (e.g. back propagation code). The code depends on many standard Python libraries used in machine learning such as scikit-learn and others.
Something I really like about Ng’s approach to teaching deep learning is the amount of theoretical background he includes. Most of the video lectures include all the mathematics in support of theory. You really get a firm understanding how a process like gradient descent works. You’ll need a strong background in math including differential Calculus, linear algebra and probability theory. He often leaves out the more mathematical proofs and suggests that people who are stronger in math might take the extra step and writing out the proof. I did this myself when the derivative of a detailed cost function was simply presented. I wanted to more fully understand how the equation was derived, so I had to crack open my favorite Calculus text to refresh my memory of calculating the derivative of logarithms.
One important feature of the courses is you’ll walk away with many useful Jupyter notebooks all ready for you to modify for other problems. It represents a great starting point so you don’t have to start coding from scratch. You also can download all of the Powerpoint course note files, text transcripts, and even the video lectures. As with other Coursera content, I like to have local copies of the videos so I can refresh my understanding of important topics like the “bias vs. variance trade-off.”
The courses give you access to lively discussion forums where you can find help with the homework, get clarifications on the lectures, and discuss trends in the industry. Although there is no official textbook for the specialization, I would highly recommend the new book “Deep Learning” by three experts in the field – Ian Goodfellow, Yoshua Bengio, and Aaron Courville.
Finally, most weeks of each course include very compelling videos of Ng interviewing various industry luminaries like Geoffrey Hinton, Yoshua Bengio, Ian Goodfellow, Yuanqing Lin, and others who will share with you their personal stories and give you career advice.
Jump Start Your Career in Deep Learning
You can enroll in each class for free but if you want certificates you’ll have to buy a Coursera subscription of $49/month and you can take as many courses you want across their entire catalog. I figure that to get the Deep Learning specialization it will cost you about $250 if you take one class at a time, an excellent value for the quality of instruction.
AI is transforming multiple industries. After finishing this specialization, you will likely find creative ways to apply it to your work. It will help you master deep learning, understand how to apply it, and build a career in AI. Highly recommended.
Contributed by Daniel D. Gutierrez, Managing Editor and Resident Data Scientist for insideAI News. In addition to being a tech journalist, Daniel also is a practicing data scientist, author, educator and sits on a number of advisory boards for various start-up companies.
Sign up for the free insideAI News newsletter.
