Two Minute Papers: How To Get Started With Machine Learning? #51

Dear Fellow Scholars, this is two-minute papers with Kato Ejolene Ifehir. I get a lot of messages from you fellow scholars that you would like to get started in machine learning and are looking for materials. Words fail to describe how great the feeling is that the series inspires many of you to start your career in research. At this point we are not only explaining the work of research scientists but creating new research scientists. Machine learning is an amazing field of research that provides us with incredible tools that help us solve problems that were previously impossible to solve. Neural networks can paint in the style of famous artists or recognize images and are capable of so many other things that it simply blows my mind. However, bear in mind that machine learning is not an easy field. This field fuses together the beauty, rigor and preciseness of mathematics with the useful applications of engineering. It is also a fast moving field on almost any given day 10 new scientific papers pop up in the repositories. For everything that I mentioned in this video there is a link in the description box and more so make sure to dive in and check them out. If you have other materials that help you understand some of the more difficult concepts, please let me know in the comments section and I’ll include them in the text below. First, some non-scientific texts to get you in the mood are recommending the road to superintelligence on a fantastic blog by the name WeightButWhy. This is a frighteningly long article for many but I guarantee that you won’t be able to stop reading it. Beware. Nick Bastrom’s superintelligence is also a fantastic read after which you’ll probably be convinced that it doesn’t make sense to work on anything else but machine learning. There’s a previous two minute papers episode on artificial superintelligence if you’re looking for a teaser for this book. Now let’s get a bit more technical with some of the better video series and courses out there. Welch Labs is an amazing YouTube channel with a very intuitive introduction to the concept of neural networks. Andrew Inc. is a chief scientist at BIDO research in deep learning. His wonderful course is widely regarded as the pinnacle of all machine learning courses and is therefore highly recommended. Nando De Freitas is a professor at the University of Oxford and has also worked with DeepMind. His course that he held at the University of British Columbia covers many of the more advanced concepts in machine learning. Regarding books, I recommend reading my favorite holy tomb of machine learning that goes by the name of pattern recognition and machine learning by Christopher Bischop. A sample chapter is available from the book if you wish to take a look. It has beautiful typesetting, lots of intuition and crystal clear presentation. Definitely worth every penny of the price. I’d like to note that I am not paid for any of the book endorsements in the series. When I recommend a book, I genuinely think that it provides great value to you fellow scholars. About software libraries, usually in most fields, the main problem is that the implementation of many state of the art techniques are severely lacking. Well, luckily in the machine learning community we have them in abundance. I’ve linked a great talk on what libraries are available and the strengths and weaknesses for each of them. At this point, you’ll probably have an idea of which direction you’re most excited about. Start searching for keywords, make sure to read the living hell out of the machine learning ready to stay up to date, and the best part is yet to come, starting to explore on your own. Thanks for watching and for your generous support, and I’ll see you next time.