Machine Learning Foundations course

On my quest to become an expert on Machine Learning, I just finished one of the numerous MOOCs out there about the topic. It is called Machine Learning Foundations: A Case Study Approach, it is taught in Coursera and it belongs to a larger 6-course ML specialization. I came accross this course reading the interesting Quora answers about how to learn ML (I personally like those from the expert Xavier Amatriain).

With that said, let’s talk a little bit about the course itself. It is an introduction to the ML topic, so don’t expect detailed and concise explanations of what the ML algorithms do (if you want some maths refer to Andrew Ng’s courses on the same Coursera platform or the extended version in Youtube). However, I think that the concepts behind usual ML topics such as regression or classification (and even deep learning!) are very well taught by the instructors. By the way, two instructors – Emily and Carlos – that achieve the perfect joke-lecture balance… :)

I liked that the course is super-practical! While I already knew most of the theoretical explanations, I found really interesting the practical python-based demos and the corresponding assignments. Being a so practical course, and due to time limitations, you can imagine that the case-study demos were based on applying algorithms already implemented in third-party libraries. I am fine with that, but my only criticism is that they use the commercial GraphLab library from Turi (whose CEO is the instructor Carlos…). They make very clear that the objective of the course is to teach Machine Learning and that this task becomes easier with a library that the instructors know very well. I completely understand that, but I personally was looking to learn ML using some more popular python packages. Anyway, if you enrol to the course you get a 1-year GraphLab license for free.

Despite all this third-party library issue, I loved the course and I learned a lot. And if you answer yes to some of these questions below, just enroll and take it:

  • Do you want to predict the value of Bill Gates house? (Regression)
  • Do you want to classify amazon comments as positive or negative? (Classification & Sentiment Analysis)
  • Do you want to find similar people to David Beckham using wikipedia text? (Document Retrieval)
  • Do you want to recommend songs based on what other users listen? (RecSys)
  • Do you want to detect if a cat is in a picture? (Deep learning)

Not decided yet? Do you feel attracted to these type of pictures? Yes? Enroll and take the course.

scatter-plot-with-matplotlib


Scatter plot of square feet vs house prices

precision-recall-curve

Precision-Recall curve analyzing the Song Recommender algorithms

tfidf-document-retrieval

TFIDF + KNN in document retrieval

roc-curve

ROC curve evaluating sentiment analysis classification

cats

Looking for cats using deep features

 

And these are the Jupyter Notebooks (iPython) that I used to complete the course.

Next stop: I will revisit Andrew Ng’s course and do the assignments in Python. Wish me luck!

  1 comment for “Machine Learning Foundations course

Leave a Reply

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