hmmm, AMD seems to have finally found a way to up the nVIDIA at the Deep Learning game
https://medium.com/intuitionmachine/building-a-50-teraflops-amd-vega-deep-learning-box-for-under-3k-ebdd60d4a93c

On reproducibility
http://clinchem.aaccjnls.org/content/early/2017/02/24/clinchem.2017.271965.long

hmm, need to refresh my postquantum crypto curriculum...
https://en.wikipedia.org/wiki/Supersingular_isogeny_key_exchange

Deep learning sometimes just means deep memory
https://arxiv.org/abs/1703.06857

More SIGGRAPH'17 awesomeness:
Inside Fluids: Clebsch Maps for Visualization and Processing
https://www.youtube.com/watch?v=xITLcT_gkYU
+paper
http://multires.caltech.edu/pubs/Clebsch.pdf

...and SIGGRAPH'16 - more applications of particle-wave duality
https://www.youtube.com/watch?v=5C9BLAXCe1I

— …I'm a full stack engineer
— Which way does the stack grow on x86?
— [blank stare]
// michaelklishin@twitter

Book promotion, level: Godlike
https://medium.com/the-mission/10-000-hours-with-claude-shannon-12-lessons-on-life-and-learning-from-a-genius-e8b9297bee8f

you can entrust government bureaucrats with your sensitive and private data, like witness protection program data, for sure!

they also get severely punished if they leak it, like half a month’s pay in fines for leaking pretty much the entire military and civilian database set (for entire EU, not only Sweden)

https://www.privateinternetaccess.com/blog/2017/07/swedish-administration-tried-glossing-leaking-eus-secure-stesta-intranet-russia/

Now this is a treat for the perfectionists' eye
> The TAC7 scanner captures and stores color, texture, gloss and other surface appearance characteristics of physical material samples.
http://www.xrite.com/categories/appearance/total-appearance-capture-ecosystem

Soft skills 2.0: talking with the machines. Teaching your fellow machines.
https://arxiv.org/abs/1707.06742

Malicious Arithmetic, mmm, tasty!
> MASCOT: Faster Malicious Arithmetic Secure Computation with Oblivious Transfer
https://eprint.iacr.org/2016/505.pdf

Augmented Medieval Books
https://medievalbooks.nl/2015/01/16/medieval-apps/

hmm, this one is unexpected, but given a second thought... mixing uncorrelated features just slightly biases them into normal distribution, keeping other properties intact, hence no wonder it works
Small Statistical Models by Random Feature Mixing '08
http://www.cs.jhu.edu/~mdredze/publications/mobile_nlp_feature_mixing.pdf

Even given that I firmly believe in the human brains' adaptability and cognitive capacities, I must admit I didn't expect it is possible for a regular person to see the world just as the bats do, via echolocation, with normal human ears.
Mind-blowing.

And you don't need to be blind to do this, either:
> ten sighted subjects were taught basic navigation skills within a few days (Wikipedia)

https://www.ted.com/talks/daniel_kish_how_i_use_sonar_to_navigate_the_world/trannoscript

A Solution of the P versus NP Problem
'nuff said.
https://arxiv.org/pdf/1708.03486.pdf

"Don't roll your own crypto". Well, don't roll your own crypto hardware as well. I was wondering why Trezor is that cheap %)
https://medium.com/@Zero404Cool/trezor-security-glitches-reveal-your-private-keys-761eeab03ff8

Not only machines confuse completely different objects. Humans too!
(lots of examples in other tweets, the link preview is not even the best one)
https://mobile.twitter.com/teenybiscuit/status/707727863571582978

Spatiotemporal Variance-Guided Filtering:
Real-Time Reconstruction for Path-Traced Global Illumination

This is huge. Literally blows every other rendering flow out of the water.
Raytraced real-time CG just became feasible for games, VR and other real-time media. Movie production becomes a lot faster, and probably cheaper, too.
http://cg.ivd.kit.edu/noscriptf.php

The referenced paper A Machine Learning Approach for Filtering Monte Carlo Noise is also interesting.
http://cvc.ucsb.edu/graphics/Papers/SIGGRAPH2015_LBF/