How do you get from Moorgate to Piccadilly Circus?
It may look easy to you (or does it?), but it’s borderline impossible for artificial intelligence. Or it was, until Google DeepMind (that employs several physicists) unleashed its Differentiable Neural Computer (DNC). According to the DeepMind scientists, neural networks are very good at recognizing patterns, or finding the best move for a given situation (as their own AlphaGo did). However, they cannot stack up a complex sequence of actions, because they cannot remember.
In a way, you could think of data as ingredients, and of the neural network as a somewhat limited cook, that can only work sequentially on them. So it can’t set aside the dough for a pie while it prepares the filling. Chef DNC can: it makes the dough, puts it in the memory, works on the filling and then finishes our data pie.
Whether we realize it or not, when we plan a subway journey we store and link information all the time. At each stop we make a decision: stay on, get down, or change. Then we remember where we are, where we started and where we’re headed. That’s how DNC do.
Big whoop—you may say—public transportation apps and satnavs have been doing this for years. Indeed they have, and are probably better at it. But they were specifically programmed for that, and that’s all they can do.
DNC wasn’t: it learned to compute paths, starting from sets of randomly connected dots (graphs, if you want to be fancy). It would memorize the different connections (what points are connected, with what line), then try to find the shortest path between two points. Time after time, it would be told whether the answer was correct.
Once it learned, DNC was able to navigate the London Tube—quite the random graph after all—, but it could just as easily guide you through the NY subway, plan out the solution to a simple puzzle, or answer complex questions about family trees.
This video shows DNC memorizing connection (who’s whose child or sibling) and answering “who is X’s maternal great uncle?” I had to take a moment to parse the question—and I’m a reasonably competent human. Try asking that to a satnav!
Often in artificial intelligence, what the machine can do is not as important as what it might learn to do. And DNC proved it can learn a lot.