Deepmind built an AI that masters Go. How did they do it? A technical introduction.
Google Deepmind announced last week that it created an AI that can play professional-level Go. The game of Go has always been something of a holy grail for game AI, given its large branching factor and the difficulty of evaluating a position.
The new AI, called AlphaGo, has already won against the current European champion in October. It is scheduled to play in March against legendary player Lee Se-dol, widely considered one of the greatest players of all time, in a tournament reminiscent of the head-to-head between IBM’s Deep Blue and Gary Kasparov.
AlphaGo is a complex AI system with five separate learning components: 3 of the components are deep neural networks built with supervised learning; one is a deep neural net built with reinforcement learning; while the final piece of the puzzle is a multi-armed bandit-like algorithm that that guides a tree search.
This might appear overwhelming, so in this post I decompose this complex AI into 5 easiy pieces to help guide the technically inclined through the paper. An excellent introduction for the layman is here.
Continue reading “5 easy pieces: How Deepmind mastered Go”
“I don’t have time” – That’s probably the most common excuse for not exercising. But what if your mind was clearer, more efficient after you hit the gym?
The majority of mortals complain bitterly of the spitefulness of Nature, because we are born for a brief span of life […]. It is not that we have a short space of time, but that we waste much of it. Life is long enough, and it has been given in sufficiently generous measure to allow the accomplishment of the very greatest things if the whole of it is well invested.
— Seneca the Younger, On the Brevity of Life
“I don’t have time” – That’s probably the most common excuse for not exercising. Sure, getting on the treadmill eats up time – time you might use to program, read a book, work on some projects, etc. But what if your mind was clearer, more efficient after you hit the gym? You’d be more productive – maybe so much more productive that the time at the gym would pay for itself.
In this article, I highlight some of the effects of aerobic exercise on the brain, and make an argument that exercise, up to very high intensity – the equivalent of an hour of jogging every day – pays for itself in increased productivity. In particular:, you’ll learn how:
- exercise – in particular aerobic exercise – improves you memory by promoting neurogenesis in the hippocampus
- exercise improves executive control, focus and mood by promoting better cerebral blood flow
- this improvement is very substantial – especially in older people
- the time you spend exercising is more than repaid in increased productivity from better focus, decision making, mood, and memory
- exercise will also increase your free time by increasing your longevity
- the effect on longevity saturates at the equivalent of an hour of jogging per day, 6 days a week
Continue reading “Exercise, be smarter, save time”
A little known feature of Zotero is that the online version can generate an RSS feed from a collection. The feed is served via https, which is fine for e.g. feedly. However, WordPress.com doesn’t like https RSS feeds, so I forwarded my feed using feedcat – a pretty terrible service, but it was the first feed burner I could find that worked with https.
The result: on the right hand column, you can now see what the papers I recently read – it updates automagically via Zotero and the internets. Pretty cool, huh?