BK’s machine learning / AI world – part 0

So machine learning, AI, deep learning etc. have been the hot topics on social media, news and every day life. AlphaGo beat the best Go player in the world, cucumber farmer use deep learning to categorize cucumbers, and Andrew Ng, former chief scientist of Baidu says that AI is the new electricity for human being. Just like how electricity has changed human life, AI has similar potential to provide great impact as well. It is very exciting to look forward what people (or should I say machine) can achieve in the recent future.

But before I start to create self driving car, smart home or image recognition and manipulation to add dog ear on your face, I need to start from the basics to know what is this all about. Fortunately we are living in this internet era, where there are a lot of free or paid resources to learn what you want to learn. Since last year, I have taken a few online courses and I would say they more or less provide different perspective to build up my foundation in this world. Below are some of the resources I have used, ranging from:

  1. https://www.coursera.org/learn/machine-learning – Machine learning course by Andrew Ng himself. It is also a Stanford University course. Though I haven’t finish the entire course.. A little bit too overwhelmed for me when I am halfway through the course. I felt that if you are very interested to know how each algorithm work in DETAIL (by that I meant the algorithm and the mathematical proof / definition behind it), it will be a very good course.
  2. Then I stumbled onto Udacity’s https://www.udacity.com/course/intro-to-machine-learning–ud120. It’s a totally different approach, as it is more on using existing library (scikit-learn) and Python programming language to work on real life problems. It gives a very good introduction on those major algorithm eg: linear regression, Naive Bayes classification, SVM, decision tree and also things that you need to know to implement a good machine learning solution (eg: data cleaning, scaling, deal with outlier etc.). Highly recommended for beginner like me 😀
  3. Then my company has company subscription to Dataquest and I quickly sign up for it. Dataquest is an interactive online platform to learn data science. If you used Codeacademy or similar platform before then it is very similar. Basically you are provided with instruction and questions, then you type in code, hit submit and see if it’s correct or not. Also quite practical as you get to work on real life datasets and you can download your own code / solution as portfolio too

Then I came across fast.ai. They provide a course on deep learning which I am also very interested in at the moment. Also from the reviews and description it is a highly practical course and it also teaches how to utilize cloud platform like AWS to run GPU instance for your deep learning code. So probably the next few post will be writing about my experience and what I have learnt from that course 🙂

While I am learning all these, I think I should start to spend a little bit of time thinking what real life problem I would want to solve using this knowledge. After all, working on real issue is the best way to enhance your knowledge, plus at the same time you are solving something pressing too. So… lets see what I can came out with while I continue my journey in the unknown world 🙂



Leave a Reply

Your email address will not be published. Required fields are marked *