Group items matching

in title, tags, annotations or url

There you go, the latest-and-greatest chip is there. Now the only remaining tiny detail - program it.

Let's buy it first and we'll figure the rest out later :P

A hybrid image classification method was employed for this specific purpose by incorporating the advantages of both pixel-based and object-based image classification approaches. This was performed in two steps: firstly, a pixel-based image classification method, i.e., artificial neural network was applied to classify potential targets with similar spectral reflectance at pixel level; and then an object-based image classification method was used to further differentiate animal targets from the surrounding landscapes through the applications of expert knowledge. As a result, the large animals in two pilot study areas were successfully detected with an average count error of 8.2%, omission error of 6.6% and commission error of 13.7%. The results of the study show for the first time that it is feasible to perform automated detection and counting of large wild animals in open savannahs from space

And Paul, it includes neural networks!

I kept telling you guys but you just laughed and laughed :))

Really nice visualizations for rectifier (ReLU) neural nets, illustrating the effects of skip-connections, depth, width, etc. on the loss function curvature.

Neural Networks to analyse sleeplessness.

Dataset automatically constructed from public images uploaded by users on FB/Instagram with hashtags used as labels! They refer to this as 'weakly supervised learning'. Then neural net fine-tuned for ImageNet and achieved record 85.4% accuracy.

Paper at http://www.nature.com/nature/journal/v548/n7669/full/nature23463.html?foxtrotcallback=true

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

The generated inputs are quite strange, I wonder where else something like this occurs.

New selection criteria?

Recently, many competitions in the computer vision domain have been won by huge convolutional networks. In the image net competition, the convolutional network approach halves the error from ~30% to ~15%! Key changes that make this happen: weight-sharing to reduce the search space, and training with a massive GPU approach. (See also the work at IDSIA: http://www.idsia.ch/~juergen/vision.html) This should please Francisco :)

...mmmmm... they use 60 million parameters and 650,000 neurons on a task that one can somehow consider easier than (say) predicting a financial crisis ... still they get 15% of errors .... reminds me of a comic we saw once ... cat http://www.sarjis.info/stripit/abstruse-goose/496/the_singularity_is_way_over_there.png

I think the ultimate solution is still to put a human brain in a jar and use it for pattern recognition. Maybe we should get a stagiaire for this..?

Memristor-Based Artificial Neural Networks, huge potential for true, high-power AI

Memristors (for memory purposes - RRAM type) on the pipeline to be launched in orbit on a cubesat http://thewhitonline.com/2016/03/news/nasa-initiative-chooses-rowan-to-launch-satellite/

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).

Researchers at Tohoku University have, for the first time, successfully demonstrated the basic operation of spintronics-based artificial neural network.

It is typically considered that brain's process of learning is highly energy efficient. While investigating how efficiently the brain can learn new information, physicists have found that, at the neuronal level, learning efficiency is ultimately limited by the laws of thermodynamics.

some biomimicry used in energy systems... maybe it sparks some ideas

not much new that has not been shared here before ... BUT: we have done relativley little on any of them. for good reasons?? don't know - maybe time to look into some of these again more closely Energy Efficiency( Termite mounds inspired regulated airflow for temperature control of large structures, preventing wasteful air conditioning and saving 10% energy.[1] Whale fins shapes informed the design of new-age wind turbine blades, with bumps/tubercles reducing drag by 30% and boosting power by 20%.[2][3][4] Stingray motion has motivated studies on this type of low-effort flapping glide, which takes advantage of the leading edge vortex, for new-age underwater robots and submarines.[5][6] Studies of microstructures found on shark skin that decrease drag and prevent accumulation of algae, barnacles, and mussels attached to their body have led to "anti-biofouling" technologies meant to address the 15% of marine vessel fuel use due to drag.[7][8][9][10] Energy Generation( Passive heliotropism exhibited by sunflowers has inspired research on a liquid crystalline elastomer and carbon nanotube system that improves the efficiency of solar panels by 10%, without using GPS and active repositioning panels to track the sun.[11][12][13] Mimicking the fluid dynamics principles utilized by schools of fish could help to optimize the arrangement of individual wind turbines in wind farms.[14] The nanoscale anti-reflection structures found on certain butterfly wings has led to a model to effectively harness solar energy.[15][16][17] Energy Storage( Inspired by the sunlight-to-energy conversion in plants, researchers are utilizing a protein in spinach to create a sort of photovoltaic cell that generates hydrogen from water (i.e. hydrogen fuel cell).[18][19] Utilizing a property of genetically-engineered viruses, specifically their ability to recognize and bind to certain materials (carbon nanotubes in this case), researchers have developed virus-based "scaffolds" that

Has this not been underway for quite some time now? Not sure if this 'new era' is coming any day soon. Thoughts?

If they want to give the computers "senses" they should also go ahead and give them a body slightly taller than humans ...and guns. So once they reach a critical level of consciousness they can really go to town... http://0-media-cdn.foolz.us/ffuuka/board/tg/image/1385/54/1385549501025.jpg

Neural networks!!! However, indeed, "senses" will not make any sense towards human-like computing without bodies that physically interact with the world. That's where most of these things are going wrong. Perception and cognition are for action. Without action coming from the machine side all these ideas simply fail.

The enormous solar plant-jointly owned by NRG Energy, BrightSource Energy and Google-opened for business today ... well yesterday, but still impressive!

impressive pictures - looking at the 2nd to last and 4th to last one, I am wondering how this distributed individually control of the mirrors works - and idea?

Machine learning obviously. Most likely neural networks :P On the other hand: http://sploid.gizmodo.com/the-worlds-largest-solar-plant-is-killing-birds-meltin-1525107821

Darn... Another field taken by NN.. paul?

Liked for veridicity! :) In the end, for science, having models would be nice in order to promote understanding. But for practical applications? Neaaah