This was in the pipeline for quite some time now. I have been waiting for his lectures on a platform such as EdX or Coursera, and the day has arrived. You can enroll and start with week 1’s lectures as they’re live now.

This course is taught by none other than Dr. Yaser S. Abu – Mostafa, whose textbook on machine learning, Learning from Data is #1 bestseller textbook (Amazon) in all categories of Computer Science. His online course has been offered earlier over here.

Teaching

Dr. Abu-Mostafa received the Clauser Prize for the most original doctoral thesis at Caltech. He received the ASCIT Teaching Awards in 1986, 1989 and 1991, the GSC Teaching Awards in 1995 and 2002, and the Richard P. Feynman prize for excellence in teaching in 1996.

Live ‘One-take’ Recordings

The lectures have been recorded from a live broadcast (including Q&A, which will let you gauge the level of CalTech students taking this course). In fact, it almost seems as though Abu Mostafa takes a direct jab at Andrew Ng’s popular Coursera MOOC by stating the obvious on his course page.

A real Caltech course, not a watered-down version

Again, while enrolling note that this is what Abu Mostafa had to say about the online course: “A Caltech course does not cater to short attention spans, and it may not provide instant gratification…[like] many MOOCs out there that are quite simple and have a ‘video game’ feel to them.” Unsurprisingly, many online students have dropped out in the past, but some of those students who “complained early on but decided to stick with the course had very flattering words to say at the end”.

Prerequisites

Basic probability

Basic matrices

Basic calculus

Some programming language/platform (I choose Python!)

If you’re looking for a challenging machine learning course, this is probably one you must take.

Through this post, I’m sharing Python code implementing the median of medians algorithm, an algorithm that resembles quickselect, differing only in the way in which the pivot is chosen, i.e, deterministically, instead of at random.

Its best case complexity is O(n) and worst case complexity O(nlog_{2}n)

I don’t have a formal education in CS, and came across this algorithm while going through Tim Roughgarden’s Coursera MOOC on the design and analysis of algorithms. Check out my implementation in Python.

I get the following output:

51
100 loops, best of 3: 2.38 ms per loop

Note that on the same input, quickselect is faster, giving us:

Deep learning became a hot topic in machine learning in the last 3-4 years (see inset below) and recently, Google released TensorFlow (a Python based deep learning toolkit) as an open source project to bring deep learning to everyone.￼

If you have wanted to get your hands dirty with TensorFlow or needed more direction with that, here’s some good news – Google is offering an open MOOC on deep learning methods using TensorFlow here. This course has been developed with Vincent Vanhoucke, Principal Scientist at Google, and technical lead in the Google Brain team. However, this is an intermediate to advanced level course and assumes you have taken a first course in machine learning, or that you are at least familiar with supervised learning methods.

Google’s overall goal in designing this course is to provide the machine learning enthusiast a rapid and direct path to solving real and interesting problems with deep learning techniques.

On January 12, 2016, Stanford University professors Trevor Hastie and Rob Tibshirani will offer the 3rd iteration of Statistical Learning, a MOOC which first began in January 2014, and has become quite a popular course among data scientists. It is a great place to learn statistical learning (machine learning) methods using the R programming language. For a quick course on R, check this out – Introduction to R Programming

Slides and videos for Statistical Learning MOOC by Hastie and Tibshirani available separately here. Slides and video tutorials related to this book by Abass Al Sharif can be downloaded here.

The course covers the following book which is available for free as a PDF copy.

Logistics and Effort:

Rough Outline of Schedule (based on last year’s course offering):

Week 1: Introduction and Overview of Statistical Learning (Chapters 1-2) Week 2: Linear Regression (Chapter 3) Week 3: Classification (Chapter 4) Week 4: Resampling Methods (Chapter 5) Week 5: Linear Model Selection and Regularization (Chapter 6) Week 6: Moving Beyond Linearity (Chapter 7) Week 7: Tree-based Methods (Chapter 8) Week 8: Support Vector Machines (Chapter 9) Week 9: Unsupervised Learning (Chapter 10)

Prerequisites: First courses in statistics, linear algebra, and computing.

MIT’s Fall 2015 iteration of 6.00.2xstarts today. After an enriching learning experience with 6.00.1x, I have great expectations from this course. As the course website mildly puts it, 6.00.2x is an introduction to using computation to understand real-world phenomena. MIT OpenCourseware (OCW) mirroring the material covered in 6.00.1x and 6.00.2x can be found here.

The course follows this book by John Guttag (who happens to be one of the instructors for this course). However, purchasing the book isn’t a necessity for this course.

One thing I loved about 6.00.1x was its dedicated Facebook group, which gave a community / classroom-peergroup feel to the course. 6.00.2x also has a Facebook group. Here’s a sneak peak:

Thesyllabus and schedule for this course is shown below. The course is spread out over 2 months which includes 7 weeks of lectures.

The prerequisites for this course are pretty much covered in this set of tutorial videos that have been created by one of the TAs for 6.00.1x. If you’ve not taken 6.00.1x in the past, you can go through these videos (running time < 1hr) to judge whether or not to go ahead with 6.00.2x.

I have finally embarked on my first machine learning MOOC / Specialization. I love Python, and this course uses Python as the language of choice. Also, the instructors assert that Python is widely used in industry, and is becoming the de facto language for data science in industry. They use IPython Notebook in their assignments and videos.

The specialization offered by the University of Washington consists of 5 courses and a capstone project spread across about 8 months (September through April). The specialization’s first iteration kicked off yesterday.

Key Learning Outcomes
– Identify potential applications of machine learning in practice.
– Describe the core differences in analyses enabled by regression, classification, and clustering.
– Select the appropriate machine learning task for a potential application.
– Apply regression, classification, clustering, retrieval, recommender systems, and deep learning.
– Represent your data as features to serve as input to machine learning models.
– Assess the model quality in terms of relevant error metrics for each task.
– Utilize a dataset to fit a model to analyze new data.
– Build an end-to-end application that uses machine learning at its core.
– Implement these techniques in Python.

Week-by-Week Week 1: Introductory welcome videos and the instructors’ views on the future of intelligent applications Week 2: Predicting House Prices (Regression) Week 3: Classification (Sentiment Analysis) Week 4: Clustering and Similarity: Retrieving Documents Week 5: Recommending Products Week 6: Deep Learning: Searching for Images

EDIT

It’s been 3 days since the course began, and here’s how the classmate demographic looks like:

I enrolled in Introduction to Computer Science and Programming Using Python with the primary objective of learning to code using Python. This course, as the name suggests, is more than just about Python. It uses Python as a tool to teach computational thinking and serves as an introduction to computer science. The fact that it is a course offered by MIT, makes it special.

As a matter of fact, this course is aimed at students with little or no prior programming experience who feel the need to understand computational approaches to problem solving. Eric Grimson is an excellent teacher (also Chancellor of MIT) and he delves into the subject matter to a surprising amount of detail.

The video lectures are based on select chapters from an excellent book by John Guttag. While the book isn’t mandatory for the course (the video lectures do a great job of explaining the material on their own), I benefited greatly from reading the textbook. There are a couple of instances where the code isn’t presented properly in the slides (typos or indentation gone wrong when pasting code to the slides), but the correct code / study material can be found in the textbook. Also, for explanations that are more in-depth, the book comes in handy.

MIT offers this course in 2 parts via edX. While 6.00.1x is is an introduction to computer science as a tool to solve real-world analytical problems, 6.00.2x is an introduction to computation in data science. For a general look and feel of the course, this OCW link may be a good starting point. It contains material including video lectures and problem sets that are closely related to 6.00.1x and 6.00.2x.

Each week’s material of 6.00.1x consists of 2 topics, followed by a Problem Set. Problem Sets account for 40% of your grade. Video lectures are followed by finger exercises that can be attempted any number of times. Finger exercises account for 10% of your grade. The Quiz (kind of like a mid-term exam) and the Final Exam account for 25% each. The course is of 8 weeks duration and covers the following topics (along with corresponding readings from John Guttag’s textbook).

From the questions posted on forums, it was apparent that the section of this course that most people found challenging, was efficiency and orders of growth – and in particular, the Big-O asymptotic notation and problems on algorithmic complexity.

Lectures on Classes, Inheritance and Object Oriented Programming (OOP) were covered really well in over 100 minutes of video time. I enjoyed the problem set that followed, requiring the student to build an Internet news filter alerting the user when it noticed a news story that matched that user’s interests.

The final week had lectures on the concept of Trees, which were done hurriedly when compared to the depth of detail the instructor had earlier gone to, while explaining concepts from previous weeks. However, this material was covered quite well in Guttag’s textbook and the code for tree search algorithms was provided for perusal as part of the courseware.

At the end of the course, there were some interesting add-on videos to tickle the curiosity of the learner on the applications of computation in diverse fields such as medicine, robotics, databases and 3D graphics.

The Wiki tab for this course (in the edX platform) is laden with useful links to complement each week of lectures. I never got around to reading those, but I’m going through them now, and they’re quite interesting. It’s a section that nerds would love to skim through.

I learnt a great deal from this course (scored well too) putting in close to 6-hours-a-week of study. It is being offered again on August 26, 2015. In the mean time, I’m keeping my eyes open for MIT’s data science course (6.00.2x) which is likely to be offered in October, in continuation to 6.00.1x.

In fact, my standard advice to graduate students these days is “go to the computer science department and take a class in machine learning.” There have been very fruitful collaborations between computer scientists and statisticians in the last decade or so, and I expect collaborations between computer scientists and econometricians will also be productive in the future.

I have just begun working on a MOOC on algorithms offered by Stanford. Since this course gives us the liberty to choose a programming language, there isn’t any code discussed in those lectures. I plan to convert any algorithm discussed in those lectures into Python code. Since Merge Sort was the first algorithm discussed, I’m starting with that.

— Recursively sort the first half of the input array. — Recursively sort the second half of the input array. — Merge two sorted sub-lists into one list.

C = output [length = n]
A = 1st sorted array [n/2]
B = 2nd sorted array [n/2]
i = 0 or 1 (depending on the programming language)
j = 0 or 1 (depending on the programming language)

for k = 1 to n

if A(i) < B(j) C(k) = A(i) i = i + 1

else if A(i) > B(j) C(k) = B(j) j = j + 1

Note: the pseudocode for the merge operation ignores the end cases.

Visualizing the algorithm can be done in 2 stages — first, the recursive splitting of the arrays, 2 each 2 at a time, and second, the merge operation.

Here’s the Python code to merge sort an array.

We can divide a list in half log_{2}n times where n is the length of the list. The second process is the merge. Each item in the list will eventually be processed and placed on the sorted list. So the merge operation which results in a list of size n requiresn operations. The result of this analysis is that log_{2}n splits, each of which costs n for a total of nlog_{2}n operations.

This class (Part 1 of a 2-part course on interactive programming using Python – and the first course of the Fundamentals of Computing Specialization offered by RICE Unviersity) was an excellent introduction to programming because of its focus on building interactive (and fun) applications with the lessons learned each week. Most introductory coding classes start with text based (boring?) programs, while all through this course you’re required to build a series of projects that get progressively complicated with every passing week. I’m not to be mistaken to be trashing conventional pedagogy, but then again, how many gifted coders do you know who learned to code after completing all the exercises, cover-to-cover of some programming textbook? The best way to learn to enjoy coding would be to build interactive stuff, and this course scores full points on that.

A short introduction to the class in a charmingly nerdy way

The mini-projects / assignments during the course are implemented on a cloud-based environment called CodeSkulptor (built by Scott Rixner, one of the instructors for this course). I found CodeSkulptor unique, in that it allows you to share your code (because it’s browser based) with just about anyone with an Internet connection and makes you work with a graphic user interface (GUI) module similar to Pygame, called Simplegui. It also had a debugging tool, called Viz Mode that helped visualize the process. It eases the task of debugging your code and you’ll realize how cool it is as you start using it more.

Since the course mini-projects were peer-reviewed, evaluating other people’s code also became a more straight-jacket affair, as everyone has their code on the same version of Python. This ensures that the focus is on learning to code, without wasting time on the logistics of programming environment (tuning differences in versions or IDEs). I especially enjoyed peer grading – for each mini project we completed, we had to evaluate and grade the work of 5 others. This was very rewarding – because I got the opportunity to fix bugs in others’ code (which makes you a better coder, I guess) and also got to see better implementations than the ones I had coded, further enriching the learning experience. Indeed, the benefits of peer grading and assessment have been well studied and documented.

Of all the assignments, the one I loved the most was implementing the classic arcade game Pong. You could try playing a version of the game I implemented here. It is a 2-player implementation, but you can play it as a single-player game, only if you imagine yourself to be answering this somewhat cheeky question! Which Pong character are you? Left or Right?

The principal reason behind my joining this course was the way it is structured and taught. We had to watch two sets of videos (part a and part b) and then complete one quiz for each set. The main task for each week was to complete a mini-project that was due along with the quizzes early Sunday morning, followed by assessment of peers’ mini-projects on the following Sunday-Wednesday. The instructors clearly put in A LOT OF WORK to make the lecture videos interesting, laced with humor, with just enough to get you going on your own with the week’s mini-project. That way you’d spend less time viewing the lecture videos, spending more time on actually getting the code for your mini-project to work. So in a way, one might say this course doesn’t follow standard pedagogy for an introductory programming course, but then, as Scott Rixner assures, “You’d know enough to be dangerous!”

– Rock Paper Scissors Lizard Spock: A simple implementation played with the computer. This project covers basics on statements, expressions and variables, functions, logic and conditionals [I’m a huge fan of The Big Bang Theory, so I was obviously eager to complete this game. Instead of a series of if-elif-else clauses, this implementation used modular logic, all of which is taught in a really fun way. A great way to start off the course].
– Guess the Number: Computer chooses a random number between 1 and 100 and you guess that number. It covered event-driven programming, local and global variables, buttons and input fields [This game although fun, might have been more interesting to code if the computer had to guess the number that the player chose, using bisection search].
– Stopwatch: This was the first project that used a graphic user interface, using some modular arithmetic to get the digits of the ticking seconds in place. A game was also built on it where the player had to stop the watch right at the start of a second to score points. This game tested your reaction-time. It covered static drawing, timers and interactive drawing.
– Pong: The last project of Part 1 and the most fun. Creating the game required only a minor step-up from learnings from previous weeks. It covered knowledge of lists, keyboard input, motion, positional/velocity control. Coding the ball physics where you put to use high-school physics knowledge of elasticity and collisions was very enjoyable. In my game, I set elasticity = 1 (for perfectly elastic collisions)

In an interview with the founders of this MOOC, who spent they say that they spent over 1000 hours building it (Part 1 and Part 2 combined, I guess). That’s an awful lot of effort and it all shows in how brilliantly the class is executed. The support system in the class is excellent. You’ll always find help available within minutes of posting your doubts and queries on the forums. I’ve seen Joe Warren (one of the main instructors of the course) replying to forum posts quite regularly. In addition, there was enough supplementary material in the form of pages on concepts and examples, practice exercises, and video content created by students from previous iterations of the class to better explain concepts and aspects of game-building, improving upon the lecture material.

Concepts and Examples

Practice Exercises

Student-created Videos Explaining Concepts

Overall, I had a great learning experience. I completed Part 1 with a 100 per cent score even though I had a minor hiccup while building the game Pong, which was the most satisfying of all the projects in Part 1. I would review Part 2 when I’m done with that in August this year. I’d easily recommend this course to anyone wishing to start off with Python. It is a great place to be introduced to Python, but it shouldn’t be your ONLY resource. I have been taking MIT’s 6.01x introductory Python course side-by-side. I shall review that course as soon as I’m through with it. That course is pedagogically more text-bookish, and indeed they do profess the use of their textbook to complement the course. I’m 4 weeks into that course and finding that enjoyable too – albeit in a different way. I still haven’t lost a point on any of the assignments or finger exercises there, and hope the trend continues:

PS: In one of the forum threads, Joe posted a list of resources that could be referred to in addition to the class.

http://pythonbooks.revolunet.com/ – about 50 books – Another good list of free python books that is kept up to date, and I believe are all free or open-source: (I won’t repeat all the books on the list here, just go check it out! Some are also on the list above, but not all)