Group items tagged

art meets science

Toward Autonomous Humanoids check out our new video with the iCub in the IM-CLeVeR project

Admit it ... You have fallen in love ....

you dont' know how often we had to shoot that scene :) but it is an adorable baby robot (if it works :))

Hard work in functional materials is driving the development of new wearable electronics that could be advantageous for communications and sensing is being pursued in the Advanced Functional Fabrics of America Institute. Through embroidering circuits into fabric with 0.1 mm precision full integration of electronic components such as sensors and computer memory devices into clothing is now possible.

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. :))

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

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

A sonification experiment taking data from ATLAS and translating it into music. The outcome was played at Montreux jazz fest, listen to the results in soundcloud https://soundcloud.com/sonification-quantizer. The way it works is "A tiny subset of collision data from the ATLAS Detector (in CERN, Switzerland) is being generated and streamed in real-time into a sonification engine built atop Python, Pure Data, Ableton, and IceCast." Code's in github https://github.com/cherston/Quantizer_public

"University of Copenhagen fired seismologist Hans Thybo, president of the European Geosciences Union. The official explanation for Thybo's dismissal - his alleged use of private e-mail for work, and telling a postdoc that it is legitimate to openly criticize university management - seems petty in the extreme."

Implementation of the deep learning-based image classifier (online). Try making a picture with your phone and upload it there. Pretty impressive results. EDIT: Okay, it works the best with well exposed simple objects (pen, mug).

All Prior Art is a project attempting to algorithmically create and publicly publish all possible new prior art, thereby making the published concepts not patent-able. The concept is to democratize ideas, provide an impetus for change in the antiquated patent system, and to preempt patent trolls. The system works by pulling text from the entire database of US issued and published (un-approved) patents and creating prior art from the patent language. While most inventions generated will be nonsensical, the cost to computationally create and publish millions of ideas is nearly zero - which allows for a higher probability of possible valid prior art.

A team of researchers in TU berlin and Caltech used Schrödinger's equation to visualise superfluidic flow (fluids with zero viscosity). Smoke and dry ice are close to this state in nature. Strangely enough for an equation that describes subatomic particles, it works, and reproduces experimental results better that Euler and Langrange schemes.

A TED talk about Google X way of working and some closer look at the project loon (stratospheric balloon network for internet connection) by the 'moonshot captain' Astro Teller.

A new more efficient type of single photon light source consisting of a quantum dot reproduces 1954 Robert Dicke theoretical proposal. Applications in quantum communications directly on the target. "All light sources work by absorbing energy - for example, from an electric current - and emit energy as light. But the energy can also be lost as heat and it is therefore important that the light sources emit the light as quickly as possible, before the energy is lost as heat."

A team of Russians crowdfunding a satellite to annoy astronomers ;) A satellite that will become the brightest star in the night sky

team? only one person, correct?

according to the website they have a whole team of people working on various aspects of the mission.

"The field of biomimicry mostly involves chemists, engineers and materials scientists. Fewer than 8% of the nearly 300 studies on biomimetics published in the past 3 months and indexed in the Thomson Reuters Web of Science had an author working in a biology department - a crude proxy for 'a biologist'."

AutoDraw is a new kind of drawing tool. It pairs machine learning with drawings from talented artists to help everyone create anything visual, fast. There's nothing to download. Nothing to pay for. And it works anywhere: smartphone, tablet, laptop, desktop, etc. AutoDraw's suggestion tool uses the same technology used in QuickDraw, to guess what you're trying to draw. Right now, it can guess hundreds of drawings and we look forward to adding more over time. If you are interested in creating drawings for others to use with AutoDraw, contact us here. We hope AutoDraw will help make drawing and creating a little more accessible and fun for everyone.

I hate scorpions, but this could be a nice subject for a future Ariadna study! This north African desert scorpion, doesn't dig burrows to protect itself from the sand-laden wind (as the other scorpions do). When the sand whips by at speeds that would strip paint away from steel, the scorpion is able to scurry off without apparent damage.

Nice research, though they have done almost all the work that we could do in an Ariadna, didnt they? "To check, they took further photographs. In particular, they used a laser scanning system to make a three-dimensional map of the armour and then plugged the result into a computer program that blasted the virtual armour with virtual sand grains at various angles of attack. This process revealed that the granules were disturbing the air flow near the skeleton's surface in ways that appeared to be reducing the erosion rate. Their model suggested that if scorpion exoskeletons were smooth, they would experience almost twice the erosion rate that they actually do. Having tried things out in a computer, the team then tried them for real. They placed samples of steel in a wind tunnel and fired grains of sand at them using compressed air. One piece of steel was smooth, but the others had grooves of different heights, widths and separations, inspired by scorpion exoskeleton, etched onto their surfaces. Each sample was exposed to the lab-generated sandstorm for five minutes and then weighed to find out how badly it had been eroded. The upshot was that the pattern most resembling scorpion armour-with grooves that were 2mm apart, 5mm wide and 4mm high-proved best able to withstand the assault. Though not as good as the computer model suggested real scorpion geometry is, such grooving nevertheless cut erosion by a fifth, compared with a smooth steel surface. The lesson for aircraft makers, Dr Han suggests, is that a little surface irregularity might help to prolong the active lives of planes and helicopters, as well as those of scorpions."

What bugs me (pardon the pun) is that the dimensions of the pattern they used were scaled up by many orders of magnitude, while "grains of sand" with which the surface was bombarded apparently were not... Not being a specialist in the field, I would nevertheless expect that the size of the surface pattern *in relation to* to size of particles used for bombarding would be crucial.

The International Space Apps Challenge is now over, it is interesting to have a look at the solutions ...

what sort of "app" is this e.g. http://spaceappschallenge.org/challenge/brightest-night/solution/132 "We started of by trying to set our camera in raw mode, howsoever this can't be done for current logitech modules and without a high risk of permanently destroying the camera (reflashing the EEPROM). we use a logitech c920 and captured a raw image from it using guvcview on ArchLinux we then wrote some python code to calculate the average BGR value of our Image. We are currently workin on automiztion, but this will take much longer as there will be a lot of reverse engineering of the USB-Protocol needed."

c'mon really! To name a few: - Tour of the Solar System ... that is one of the numerous specialities of the ACT. - Satellite data correlation tool ... solution is flawed, but I really like the idea. Lot of added value. I definitely think are interestung stuffs to do there for cheap. - Aurora project: Model & Data ... space weather not interesting?! - Kepler ... a classic Do not expect the solutions to be tremendous just after 3 days of work, but the proposed list of challenges are "seeds" for innovation to me.

It is not new, but I can't remember any discussion on this concept at the ACT

"It works by emitting a burst of light from a femtosecond laser that reflects off visible surfaces - such as an opaque wall - onto objects that are hidden from the camera's direct view. The light then bounces off the object before ultimately making its way back to a detector. This process is repeated a number of times with the laser targeted at different areas of the reflecting surface." nice video