My colleague, Andrew Walker, writes about his experience doing Stanford University’s AI course in a guest (and his first) blog post.
In August last year, tech enthusiasts were buzzing with the news that Stanford University would be offering three classes online and free, in the upcoming spring semester. The classes being offered included an Introduction to Artificial Intelligence, Introduction to Databases and Introduction to Machine Learning. This buzz grew so loud that mainstream media such as the New York Times picked up on it. The amount of attention it gained was amazing and at first glance hard to comprehend as Universities like Stanford had been making course material available online for the general public for quite a while through facilities like iTunes U. However Stanford was taking this a step further, these where going to be “real” classes with homework assignments deadlines and examinations in the advanced stream.
I was one of 160 000 people who signed up for the Artificial Intelligence class. My peers came from 190 countries, spoke 40 different languages and although I don’t know for sure our youngest classmates may have been the 11 year old twins that mentioned their age whilst discussing mid term exam scores and the oldest could well have been the person who confessed to being in their late 80s.
Communities quickly developed on Reddit, Aquis and Facebook. Volunteers stepped forward to translate the lessons from English for the many no English speakers. These communities were one of the major differences between this class and the other lecture material that I have worked my way through online. It was reassuring to log onto Aquis and see people seeking clarification of the same points I had had to ponder and to see the responses of the many people who offered help.
The class consisted of a weekly lesson with associated homework task, mid term exam and final exam. Each lesson was broken down into sections with each section starting with a video presentation from either Sebastian Thurn or Peter Norvig followed by a quiz. After attempting the quiz you could choose to watch a video explanation of each question or move onto the next section of the lesson. I found being able to rewatch a section of the lesson before moving on invaluable and a vast improvement on what I experienced sitting in lecture theatres many years ago, where missing a concept could make all of the material that followed meaningless. I suspect that the traditional lectures for those physically attending classes in many universities will soon be replaced by this prerecorded material that students watch and lecturers time spent in more interaction with students who have already viewed the material.
The first lesson I found straightforward and the homework assignment relatively easy. The second lesson made it obvious that my maths just wasn’t at the level required to do the work, so it was off to the Khan Academy for crash courses in linear algebra and probability before attempting the homework.
After completing the advanced track I am proud to say I am one of the 23 000 who passed. I definitely learnt a lot, not just about the areas of Artificial Intelligence covered in the class but also in the mathematics and logic underpinning them. The homework deadlines stopped me from putting off doing lessons indefinitely which has been one of my problems when trying to work through other course material by myself. The online communities at Aquis and Reddit provided reassurance when it was obvious from discussions that I was not the only one finding the going difficult at times.
This class was a very worthwhile experience and I have signed up for Introduction to Machine Learning and Design and Analysis of Algorithms 1 from Stanford and also CS 101: Building a Search Engine from Udacity beginning mid February.
Does anyone else have experiences to share from these courses?