Showing posts with label Coursera. Show all posts
Showing posts with label Coursera. Show all posts

Tuesday, May 01, 2012

An attempt at a Cape Town catch up

Time simply doesn't allow to do all things I want to do, and to write about them. As of the last post my life has been taken over by the online courses I mentioned, and this means working till 2 most mornings to get through the material while having a normal time in the department during the day. I wrote the following when waiting for my plane in Cape Town, and while it's by no means a polished blog post, I want to post it before I disappear again tomorrow morning for Copenhagen. The next 3 days are part of an absurd trip which has been in the planning for several years and will come to its climactic conclusion on Thursday with lunch at NOMA.

-----------

I find myself once again sat in a cafe, in an airport, waiting for another 12+ hours of plane flights. This time I'm in Cape Town, waiting to go via Johannesburg back to Munich. Another overnight flight, another night of watching others loll in and out of sleep as I wander the aisles trying to get tired enough to ignore the discomfort of folding my frame into a spasm-inducing gap.

Anyway, the above sounds a bit negative perhaps and the rest of the post, I hope will not be. I've met enough interesting people on flights now to know that flights have the possibility to be as enjoyable as they are ghastly.

I've spent the last three weeks in Cape Town, my second time here, and each extra day I spend here makes me realise even more what a wonderful place it is. Scenically it is almost unrivalled for a city. With the sea surrounding it and the mountain defining its geography, you are never away from a spectacular view. I've been for a couple of sundowners (a phrase I'd not heard before but means, unsurprisingly, drinks drunk during sunset, generally with a majestic view to accompany them) which have been wonderful evening highlights, a time to really sit back and appreciate.


The city itself is made up of an incredible diversity of populations and cultures, the mixing of which is often not present but always available. I've met some incredible people here, in taxis, in shops, in cafes, from refugees of Rwanda to ex-militants who fought in Umkhonto we Sizwe, the spear of the nation, the group that was the force behind the ANC when talks and peaceful action only managed to tighten the grip of Apartheid. Everyone here seems to have a story to tell and I've met so many people who are actively pushing to make the situation better. In many ways South Africa has come a long way since Apartheid, but in others it has gotten nowhere. There are many embarrassments of the current social and political situation but there are people out there who are really trying to make a difference in a non political context. One of the taxi drivers I met also worked as a coast guard guarding the local ecology from abalone poachers, who often come armed with guns during the night. This is a man who puts himself in a place of enormous danger because he cares - lessons to take from this, for sure.

I am in no way qualified to make judgement calls on the current situation but I can recommend these books that I've read over the last couple of years to give some picture of South Africa's history and recent developments. Sadly time does not allow for in depth reviews of any of them but I can say that from the point of view of someone who knew very little about Africa or its politics, these are great starters (Thanks to Ben for all of these recommendations):

The State of Africa - a history of the states of Africa since independence
Country of my skull - a journalist's eye view of the truth and reconciliation commission
My traitor's heart - a white man with a difficult family pedigree coming to terms with his place in Africa
The shackled continent (don't read this without reading The State of Africa as it will give you a terribly biased view) - a view of why things have not improved since independence. As I say, this does not give anywhere near the whole story and for the reason behind these reasons you have to read more.
Long walk to Freedom - Mandela's pre-freedom autobiography.
The autobiography of Ghandi also gives some very interesting insights into the relatively recent history of South Africa.

The trip has been as eclectic a mix as the city is and I've been both giving seminars as well as starting some collaborations with the string theory group. I gave my atmospheric optics talk to the undergraduates which is always a pleasure, though I still struggle at times not to include too much material in this talk. After this I gave a series of lectures on a particular method of machine learning as applied to neuroscience. This was my first time lecturing on a non-physics subject and it was a great experience. The subject is fascinating and thankfully there were a huge number of questions which I find always make a lecture course a lot more enjoyable. It also makes me, the lecturer, understand the topic much better when you have a lot of intelligent people giving you questions that perhaps you would never have considered.

Some light photo relief:
 The outrageously beautiful University of Cape Town campus
Ivydene guest house, where I would sit and work at the weekends under the tree with the dogs and cats resting in the garden. This is without a doubt my favourite guest house having stayed at many over the last decade or so. Lucille, who runs it is absolutely fascinating and introduced me to a slew of amazing people while I was there. Her daughter Jackie and son-in-law Rob also make the place feel like family.

And finally I gave a couple of talks about one of my last string theory papers which again was a lot of fun. The group ranges from those working in emergent geometry (how space and time can emerge from some other theory which don't include space and time as input) to AdS/QGP - how we can use string theory as a tool to understand the internal constituents of the nucleus in particular conditions. It's always good to have a diverse audience like this because, again, you get a great range of questions, from the abstract to the very practical. The outcome of this was some really nice ideas for future investigations which I hope to continue remotely.

and apart from a little socialising my life has been almost entirely overtaken with the online courses that I discussed in the previous posts. As an information addict these are truly fatal. How can I turn down a perfectly good course on automata theory given by a world expert, even if it means sleeping a few hours less a week.

The first round of courses on computer science finished a couple of weeks back which I can highly recommend, CS101 for those without much experience in programming and CS373 for the more advanced and those who may be interested in control theory and dynamical programming.

The course on natural language processing (NLP) is now over half way through and is the best taught and most instructive from the point of view of programming assignments and problem sets, of the Coursera courses I've been taking. Right now I'm struggling, along with most of the rest of the class to write our own English language parser (how do you write a computer program which can work out the logical structure of a sentence - ie. which nouns do which verbs and prepositions relate to - in the sentence "fish people fish tanks" does the word people correspond to a verb or a noun - does the second 'fish' correspond to a very or a noun, etc. These are the sorts of question that an automatic parser would try to answer in a probabilistic fashion (ie. it will find various ways to break down a sentence and give probabilities for each possibility)).

The NLP course also ties in very nicely with material in the automata course and the programming languages course from Udacity. The link being finite state machines and Markov networks.

Probabilistic Graphical Models is also very good, but it really does need 10-15 hours a week to truly understand the material. This is a large investment and something that I will probably split over two runs through the material.

In other news, the publication I was working on about Malaria has now been accepted for publication which is great to have finished. More work is ahead on this front and there are some very exciting possibilities in the pipeline. Through the same link that gave me the Malaria connection (many thanks Victor!) we've started also on another investigation into an area called medical ethnography - a subject whose existence I was unaware of until I dove into some data analysis on a related project. This area is related to the interaction between society, ethnological behaviour and medical practice and is so far very interesting to work on.

So, with the above I have very little time for a normal life right now. I work until late at night including at weekends on these projects and courses, which is great in terms of learning and in terms of the excitement of pursuing new areas of investigation but I am aware that this is not a sustainable pace. I am also aware that I am out of a job in 7 months and am desperately trying to mold a future which will be interesting, sustainable, economically viable and will allow for some constancy in my life. Most of these factors are falling into place, it's just the small matter of economics which may be the biggest hurdle. It's easy to find people with whom I can work on interesting projects, it's not so easy to get them to pay me for it. The thinking cap is on and I have a few ideas as to how to construct this Rube-Goldberg machine of a life out of gaffer tape and fuzzy felt.

Wednesday, March 14, 2012

2012 - bring on the learning revolution

(I should note that the title is taken straight from Ken Robinson's TED talk)

I'm exhausted at the moment, but hugely excited. I've been spending evenings over the last three weeks, usually till the early hours, experimenting with a new development in online education.

The Open Courseware (OCW) movement started in 1999, in Tubingen, Germany, but quickly spread to the US. The idea is simple: Take the best educators at the top universities around the world, video their courses and put them online, for free, for anyone to watch. Over the last couple of years I've watched probably a dozen or so courses on subjects from human behavioural biology, to quantum field theory (admittedly this is a little late, being Sidney Coleman, who passed away before the videos were put online), to existentialist philosophy, from science and cooking to biology at MIT. Some of these may not fall under the umbrella of OCW officially, but the idea is the same: As of the last decade, there have been a wealth of amazing materials online for anyone to access. Open Culture keeps a pretty good track of the best of the courses online here.

These courses were often fantastic to watch and I learned a great deal, but there was always a large piece missing from these sets of audio or video materials, which is that they were very much uni-directional. You sat, as a viewer, and took in the information. It turns out that this sort of learning is pretty inefficient. Building up passive understanding of a subject is one thing, but building up a true working understanding of it takes exercise and effort. The act of letting it wash over you is not enough. In fact even going through the material yourself in 'revision' mode is pretty ineffective (there's a lovely paper here detailing precisely the effects of various types of learning methods).

So, come the end of 2011 things started to change. Two courses were offered online from Stanford whereby not only were there video lectures, but there were online exercises in the form of multiple choice problem sets and programming assignments, all of which were marked automatically. Suddenly you were forced to understand the material (though in the case of the hugely successful Machine Learning course I'd say that, given that this was a first run, the tests didn't actually examine how much you understood the material, but more the programing language - in this case Octave - this will change in future revisions I'm sure). The other course, on Artificial Intelligence, gained over 100,000 participants, and suddenly the whole thing exploded. Sebastian Thrun said that, having lectured to so many students, lecturing to a classroom full of 200 enormously lucky individuals no longer seemed terribly efficient (I'm misquoting and paraphrasing enormously) and from this vision of the future, Udacity was born. Some of the quotes can be found in the article here which gives more outline of his vision.

Udacity is one of several platforms which have started in the last few of weeks. Udacity aims, eventually, to give the material of an entire degree course in computer science. The modules are not coming online in a linear fashion but this means that whatever your knowledge of programing there will probably be something for you.

I'm taking part in the first two courses. CS101 teaches you to build your own search engine, and is really a very basic course in Python with some simple implementations of web crawling and web searching but for a beginner in Python it's perfect.

CS373 is a more advanced course and expects that you already have a good working knowledge of Python, although you could pick up what you need to know for this in a matter of a few hours. The course is based around programing a robotic car - like the car that Google used to win the Darpa Grand Challenge (a challenge whereby an automated car has to make its way through long and complex real-world road situations). In this course you learn a great deal about robotics and designing intelligent computer systems for navigating complex environments.

Udacity is going to be offering more courses starting in April. Each course, I believe, will last on the order of a couple of months and at the end you get a certificate based on the marks you get in your homework assignments. Typically there are an hour or so of lectures a week and the lectures themselves are built around solving problems for yourself - you learn as you move through the lectures by solving the puzzles as you go. Here is a screenshot from one particular programming assignment in the course - not homework, but just a question that is given as you move through the second unit. This is all done online and you program directly into the browser.
You can see above the format of the class whereby not only are there lectures and homework, but also a hugely active discussion area where students discuss the course, pose questions and build understanding together.

So, that's Udacity. At the same time, another organisation, Coursera has come online and is offering 15 or so courses on everything from Natural Language Processing to Anatomy.
Right now I'm taking part in Model Thinking, Probabilistic Graphical Models, Natural Language Processing and Design and Analysis of Algorithms 1. Each of these have several hours of lectures a week, so I'm downloading them and watching them with VLC at a faster speed, slowing down for the more complex ideas. The format is very similar to the Udacity courses, though the lectures are less active for the watcher on Coursera. Again, the emphasis is on plenty of exercises to make sure that you really build understanding as you go through the course. Here's a screenshot of the Natural Language Processing page:
Here's an introduction to the model thinking course:

and on probabilistic graphical models:

In addition to all of this, MIT, through MITx is offering their own format for an online course starting with MITx 6.002 Introduction to circuits and electronics. This really astounded me. Not only do you get the same extremely high quality video lectures, homeworks, discussion forums etc. but there is, built into the interface, a virtual electronics lab for you to experiment with and use for the assignments. For completeness I'll include a screenshot of this interface as well:
They also have an ever-evolving wiki page which will allow the students to write their own online version of the course which will evolve with ever-better explanations of the concepts discussed.

What is immensely important with respect to all of these online courses is that every one of them is completely free!

As of this year, and, I predict, in vastly increasing numbers, you can now get some of the highest quality education in the world, online, for free. Anyone, in any country, with an internet connection, can register, log in and learn with some of the best educators in the world. This is mind-blowingly powerful stuff! I'm hugely excited to see what this will do for the younger generation of students (high school included) who are dying to get immersed in all this stuff but up until now simply haven't had the resources (not only the material, but also the interaction that this will now allow).

I should add incidentally that TED is also getting in on the act, having launched TED Education this week:


A discussion of online free learning would not be complete (and this blogpost is certainly not that) without a mention of Khan Academy, another incredibly exciting venture which looks to have the potential to revolutionise highschool education:


Anyway, I highly advise delving into these courses and even if something doesn't pop out now that appeals, I think that within a very short space of time, the vast library of online, interactive information available will have something to tickle your brainbuds.

So, for now, I have a little more to watch and work through tonight (for the last weeks I've been working through from about 8-1 or 2am at home, after work). I've taken on a pretty big load with all of these courses and am not aiming at perfection but for now just exploring the landscape of possibilities, so far it's a lot of fun...

P.S. I've shared this before, but this is always worth putting out there for anyone who hasn't already seen it. Sir Ken Robinson is one of the most wonderful rhetoricists I've ever seen, and I've had the pleasure of seeing him live at a conference in London. Here describes how the current education system is badly broken and needs a massive rethink. Is education killing creativity?:


Also, find out more here at Learning Without Frontiers, about how disruptive technology may be able to shift the status quo.