Learning AI and Machine Learning. Lessons from Andrew Ng
One of the questions I get asked a lot as a speaker and trainer in AI, machine learning and data science, is how and where to learn AI and machine learning. A lot of practitioners who are entering this field start by taking MOOC’s (massive open online courses) on platforms like Coursera, EdX or Udemy. Andrew Ng is perhaps the most famous lecturer on machine learning and deep learning. Millions of students have taken his online classes on Coursera and deeplearning.ai. Lex Fridman, the host of the excellent Artificial Intelligence podcast and researcher at MIT, has interviewed Andrew in episode #73 of this podcast and in this blogpost I would like to draw some lessons from this interview. You can listen to full podcast on Youtube or via your favourite podcast app.
Teaching and learning
Andrew is a renowned teacher and one of his favourite ways to explain difficult concepts is to use a whiteboard. One of the benefits of using a whiteboard over slides is that it forces the educator to really focus on the core concepts and it helps to explain complex concepts by building up your story. As a teacher I also prefer using a whiteboard to explain AI and Machine Learning concepts as you connect and interact with the students. You can see Andrew in action using a whiteboard on this video. It is definitely worth checking out.
Andrew als has some good recommendations for anyone who wants to learn AI and machine learning. The key success factor to learning is setting time apart to learn on a regular basis. Try to block e.g. 30 minutes each day and focus on the tasks so you keep making progress. It is important to develop a convenient learning discipline.
Another tip Andrew gave was to take notes using pen and paper. If you are like me always taking notes on a laptop, this feels like going back in time. There is a good rationale behind this thinking which is backed up by a vast amount of research, that taking notes using pen and paper forces you to slow down your brains and summarise the content in your own words, rather than writing down verbatim what the instructor says. A good tip would be to digitize your notes so you have them available in digital form; this is especially useful if you use software that uses optical character recognition (OCR).
How to become a data scientist
Andrew has taught millions of people who want to enter the field of machine learning and data science. His answer to the question “how can I become a (professional) data scientist?” is summarised in the following key points.
- Take as many courses as you can and be interactive using the course materials. It is also important, as mentioned above, to study on a regular basis in order to keep the pace and maximize the learning effect. Many students tend to drop out and don’t finish the course.
- Practice, practice, practice. You can only learn data science by doing. Start playing around with the course materials and data sets and once you master the basics, try to engage in e.g. a Kaggle competition or find yourself a pet project. If you are e.g. into running then use your running data to develop a machine learning model. If you are already working in an organization try to find a small project in your workplace to work on.
- Build a portfolio. Try to engage in as many projects as possible and build up your portfolio on e.g. Github.
- Blog. Start communicating about the progress you are making so you can share your experiences with the world. This will also help you to build up your portfolio
Operationalizing data science
One of the companies that Andrew has co-founded is Landing.ai. This company helps organizations to progress in their AI and data science strategy. One of the critical success factors for this is to get the machine learning model from the developers laptop into production. Andrew gave a few tips for companies who want to get to the next level of AI.
- Don’t get carried away with new and hot technologies like deep learning for technology sake. Always remember that you are trying to solve a business problem so it is important to focus on outcomes rather than technologies.
- As Covey already mentioned in his bestseller book The Seven Habit of highly effective people: always start with the end in mind. There is a huge difference between running a machine learnig model on your laptop vs. running it in production. You have to think about topics like scalability, security and data availability very early on in the machine learning development process
- Develop a clear AI strategy. Andrew’s free AI Cookbook is an excellent resource for anyone who wants to lead their company in the AI era.
All in all I would really encourage you to listen to this podcast as it is worth every minute. Also if you are interested in learning AI/ machine learning/ data science check out our Training page or reach out to Jack Esselink to discuss how Studio Pulpit can help in educating you or your organization.