Group items matching

in title, tags, annotations or url

Babies learn about the world by exploring how their bodies move in space, grabbing toys, pushing things off tables and by watching and imitating what adults are doing. But when roboticists want to teach a robot how to do a task, they typically either write code or physically move a robot's arm or body to show it how to perform an action.

Not new but still crazy: neutrinos explored as a source of some mass extinctions.

The Sun's gradual brightening will seriously compromise the Earth's biosphere within ~ 1E9 years. If Earth's orbit migrates outward, however, the biosphere could remain intact over the entire main-sequence lifetime of the Sun. In this paper, we explore the feasibility of engineering such a migration over a long time period. (via Nina)

any papers or insights on methods available?

Couldn't find a paper for Alibaba's results but Microsoft Research's performance on this dataset was very close. The paper is here: https://www.microsoft.com/en-us/research/wp-content/uploads/2017/05/r-net.pdf Btw the 'reading test' is a publicly available dataset called 'Stanford Question Answering Dataset (SQuAD)'. Their website shows a leaderboard: https://rajpurkar.github.io/SQuAD-explorer/

interns at NASA are looking into soft robot actuators

first science results reported in Nature (as far as I know) from the Yutu-2 and Chang'e mission .... and they look very good!

Sure they are very useful! It will be even better if they manage to fit the data to modeled circulation of the lunar magma ocean that was formed posterior to the "Theia" body collision with Earth. The collision was the cause of the magma ocean in the first place. The question now is how this circulation pattern of the lava-moon "froze" in time upon phase transition to solid. Because, what crystallizes last in sequence, is more rich in "incompatible" with the crystal structure, elements, we might combine data+models to predict their location. Those incompatible tracers are mainly radioactively decaying elements that produce heat (google publications about lunar KREEP elements (potassium (K), rare earth elements(REE), and phosphorus(P)). By knowing where the KREEP is: - we know where to dig for them mining (if they are useful for something, eg. Phosphorus for plants to be grown on the Moon) - we avoid planning to build the future human colony on top of radioactives, of course. The hope is that the Moon, due to lack of plate tectonics, has preserved this "signature of the freezing sequence". Let's see.

thanks Nasia! very interesting comment

Some locations on Mars are known to have water ice just below the surface, but how much has remained unclear. Dundas et al. used data from two orbiting spacecraft to examine eight locations where erosion has occurred. This revealed cliffs composed mostly of water ice, which is slowly sublimating as it is exposed to the atmosphere. The ice sheets extend from just below the surface to a depth of 100 meters or more and appear to contain distinct layers, which could preserve a record of Mars' past climate. They might even be a useful source of water for future human exploration of the red planet.

nice, did any of you try it already

Recent study on a moon-based space elevator towards GEO

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

Might have some interesting datasets.

Couldn't find more details on this but sounds interesting.

interesting indeed as a concept and to explore how far one can go to overcome some of the natural deficiencies of wood. regarding the main purpose of debris avoidance, it's typically not the structure that survives re-entry and one would have to see the environmental effect of what would be released in the atmosphere when burning.