Group items tagged

"Leading scientists have drawn up plans for a vast multinational European institute devoted to world-class artificial intelligence (AI) research in a desperate bid to nurture and retain top talent in Europe. The new institute would be set up for similar reasons as Cern, the particle physics lab near Geneva, which was created after the second world war to rebuild European physics and reverse the brain drain of the brightest and best scientists to the US."

An interesting project that can teach you something about AI training bias. While the system can generate marvelous portraits, it was trained on images displaying not a smiling expression, so it will most likely fail on smiling photos.

This has been around for over a year. The current trend in deep learning is "deeper is better". But a consequence of this is that for a given network depth, we can only feasibly evaluate a tiny fraction of the "search space" of NN architectures. The current approach to choosing a network architecture is to iteratively add more layers/units and keeping the architecture which gives an increase in the accuracy on some held-out data set i.e. we have the following information: {NN, accuracy}. Clearly, this process can be automated by using the accuracy as a 'signal' to a learning algorithm. The novelty in this work is they use reinforcement learning with a recurrent neural network controller which is trained by a policy gradient - a gradient-based method. Previously, evolutionary algorithms would typically be used. In summary, yes, the results are impressive - BUT this was only possible because they had access to Google's resources. An evolutionary approach would probably end up with the same architecture - it would just take longer. This is part of a broader research area in deep learning called 'meta-learning' which seeks to automate all aspects of neural network training.

Btw that techxplore article was cringing to read - if interested read this article instead: https://research.googleblog.com/2017/05/using-machine-learning-to-explore.html

smart students ...

... stupid AI

In the paper authors Daniel Hennes (former RF in AI here at the ACT) ....

Freud and AI ...

reads way too pessimistic to me - or too optimistic if you are an AI guy .... Luis? Dario?

What makes us different is the particulars of our history, which gives us our notions of purpose and goals. That's a long way of saying when we have the box on the desk that thinks as well as any brain does, the thing it doesn't have, intrinsically, is the goals and purposes that we have. Those are defined by our particulars-our particular biology, our particular psychology, our particular cultural history. The thing we have to think about as we think about the future of these things is the goals. That's what humans contribute, that's what our civilization contributes-execution of those goals; that's what we can increasingly automate.

Get cracking, ACT AI team ;) A new X prize designed to advance artificial intelligence has been announced at the TED (Technology, Entertainment and Design) conference. The X Prize was set up to push the boundaries of technology to solve issues such as climate change. The winner, which will be announced at TED in 2020, will win $5m (£3.4m).

For all the AI guys (Christos, Marek, Ed, Markus and co ...) and of course Luiz, Sante ... You will like this one :-)

Quite a lot of blabla, some usual misconceptions (like QT the source of randomness in nature), but a -- from my point of view -- very true (though in the text somehow hidden) conclusion: Free will, creativity etc. from the point of view of fundamental physics are just randomness! Many physicists won't like this conclusion, though, and in this respect also the title is rather misleading!

I have no idea if an algorithm based on this already exists, but it would certainly be a good one for autonomous AI, I think. I think an algorithm based on this should be able to select his own input parameters and reject them if they are not stimulated any further or integrate them in the algorithm if they are continiously stimulated... this could enable self learning, etc.

By steering the neuron's back to an intermediate activity level the mechanism probably optimizes their efficiency within the network (after all a neuron that fires all the time is just as useless as one that never fires).

A bit of self-ad here :-) Hear my lab colleague Tony Morse speaking about developmental robotics, and meet our little iCub... As a bonus, have a peep into the kitchen and messy lab of the guys downstairs... my office is of course much nicer!!!

"and yet one (idea) that allowed to make new predictions that are now being tested in children"!!! Which one? I am curious.

Will ask him today...

Of any interest to our AI gurus?

okay this is posted now 3 times ;) congrats you were the first :)

for our AI guys ...

nice article (in italian, sorry) that made me think that we should seek for artificial creativity rather than AI

Explanation of artificial neural networks (explains very nicely the basics of the programming behind Christos' PhD thesis)

The best post of Slashdot: "Genetic Algorithms are like the AI equivalent of text editors... everybody has spent a weekend writing one at some point."

semantic web is here as well as AI and other ACT related ideas

Here we go! AI for Mario has arrived. What is still missing is the Italian accent in his robo-voice though. To compensate for that, we have a lot of German accent in the Youtube-video. Make sure to check it out if you always liked the background music from the games. Putting the trolling aside, I honestly like the idea! There is still some way to go, but maybe we will watch and even pay to be able to see twitch-streams of self-ware bots one day?

I'm actually considering doing an twitch-bot channel. Why pay when you can cash in? Kind of lazy tho`

Let an AI develop your magic tricks! (The one with the smart phone is actually neat)

They published it in Frontiers of Psychology...?