Advanced Concepts Team

Cool hologram generating tabletop. Check the videos. Would be interesting if it could be scaled down..

A free software tool that could revolutionize the drone industry to design and build your own drone/aircraft/fighter with a 3-D online software that generates aerodynamic information (maybe optimization also) about your own design.

Amazing engineering feat: 1 nm transistor. Besides we can argue Moore law is still OK, dennard scaling is gone and with it the performance boost, as alluded subtly. Link article: http://science.sciencemag.org/content/354/6308/99.full

IBM for the first time ever is making quantum computing available to the public, providing access to a quantum processor via the cloud. Users can create algorithms and run experiments and get inspired by the possibilities of a quantum computer.

Looks interesting.. Have you tried it?

Mathias Troyer from ETH Zurtich gave a talk in Leiden where he showed what he wants to be the replacement to this IBM programming or the best ally of it - program quantum computers with, for instance, python code. Nice developments coming from the quantum coding field, besides the fact we are ages away from a practical quantum computer.

A good historical overview of the Deep Learning revolution. If you think the quote above is an exageration, here are some fresh news from Microsoft: Internal email: Microsoft forms new 5,000-person AI division

"IBM delivered on the DARPA SyNAPSE project with a one million neuron brain-inspired processor. The chip consumes merely 70 milliwatts, and is capable of 46 billion synaptic operations per second, per watt-literally a synaptic supercomputer in your palm." --- No memristors..., yet.: https://www.technologyreview.com/s/537211/a-better-way-to-build-brain-inspired-chips/

Physicists from MIT explain why deep learning is that much successful using physics.

Very interesting presentation on how the brain can back-propagate error signals during learning (using time-derivatives to encode errors). Hinton discusses how back-propagation can be achieved with very limited / unsophisticated tools and in excessively noise environments.

from one of the reddit threads discussing this: "bit fishy, crazy if real". "Incredible claims: - Train only using about 10% of imagenet-12, i.e. around 120k images (i.e. they use 6k images per arm) - get to the same or better accuracy as the equivalent VGG net - Training is not via backprop but more simpler PCA + Sparsity regime (see section 4.1), shouldn't take more than 10 hours just on CPU probably "

clicking the link says the manuscript was withdrawn :))

This "one-shot learning" paper by Googe Deepmind also claims to be able to learn from very few training data. Thought it might be interesting for you guys: https://arxiv.org/pdf/1605.06065v1.pdf

Cute info-graphics on different machine learning architectures

I got a fever. And the only prescription is more cat faces! ...../\_/\ ...(=^_^) ..\\(___) The article sounds quite interesting, though. I think the idea of a "fake" agent that tries to trick the classifier while both co-evolve is nice as it allows the classifier to first cope with the lower order complexity of the problem. As the fake agent mimics the real agent better and better the classifier has time to add complexity to itself instead of trying to do it all at once. It would be interesting if this is later reflected in the neural nets structure, i.e. having core regions that deal with lower order approximation / classification and peripheral regions (added at a later stage) that deal with nuances as they become apparent. Also this approach will develop not just a classifier for agent behavior but at the same time a model of the same. The later may be useful in itself and might in same cases be the actual goal of the "researcher". I suspect, however, that the problem of producing / evolving the "fake agent" model might in most case be at least as hard as producing a working classifier...

This paper from 2014 seems discribe something pretty similar (except for not using physical robots, etc...): https://papers.nips.cc/paper/5423-generative-adversarial-nets.pdf

Yes, this IS basically adversarial learning. Except the generator part instead of being a neural net is some kind of swarm parametrization. I just love how they rebranded it, though. :))

Some interesting reads 150 years after H. G. Wells and 50 years after Star Trek.

"The One Hundred Year Study on Artificial Intelligence, launched in the fall of 2014, is a long-term investigation of the field of Artificial Intelligence (AI) and its influences on people, their communities, and society." (...) "The report is designed to address four intended audiences. For the general public, it aims to provide an accessible, scientifically and technologically accurate portrayal of the current state of AI and its potential. For industry, the report describes relevant technologies and legal and ethical challenges, and may help guide resource allocation. The report is also directed to local, national, and international governments to help them better plan for AI in governance. Finally, the report can help AI researchers, as well as their institutions and funders, to set priorities and consider the ethical and legal issues raised by AI research and its applications."

google paper on quantum computers ... anybody with further insight on how realistic this is

Not an answer to Leopold's question but here is a little primer on quantum computers for those that are (like me) still confused about what they actually do: http://www.dwavesys.com/tutorials/background-reading-series/quantum-computing-primer It give a good intuitive idea of the kinds of problems that an adiabatic quantum computer can tackle, an easy analogy of the computation and an explanation of how this get set up in the computer. Also, there is emphasis on how and why quantum computers lend themselves to machine learning (and maybe trajectory optimization??? - ;-) ).

Any unintended consequences..? https://i.stack.imgur.com/A0UZn.jpg

In a world increasingly dominated by the QWERTY keyboard, UCL computer scientists have developed software which may spark the comeback of the handwritten word by analysing the handwriting of any individual and accurately replicating it.

Similar work has been around for several years: http://www.cs.toronto.edu/~graves/handwriting.html (as mentioned in the video). I think this could be nicely applied for procedurally generated handwritten post cards (e.g. invitation, thank you cards for weddings, funerals, child birth, etc.) where the sender would otherwise have to write dozens of cards. Anybody interested to start a business? ;-P

Time to get out the Fission Fragment Rocket Engine!