danielmorozoff

Historically there have been many countries / communities that have used local currencies and private issues of money to enable spending.

https://en.m.wikipedia.org/wiki/Local_currency

This was especially common during times of economic hardships: wars and famines. Ex. Russian revolution, the pogroms. Things used were food vouchers, lottery tickets, checks and obligations all sorts of things.

I wouldn't be so quick to say insurance companies would pick up the tab. I don't know which country you're from but in the us getting insurance companies to pay for a claim is very hard and until a set of laws are established I would expect dramatic pushback from them. Furthermore in most accidents insurance companies try to seek out the party at fault if it's not crystal clear to avoid paying.

But an interesting situation would occur if the incentives from the insurance companies would be directed towards automated driving because of reduced risks and costs to the driver. That would be an interesting scenario

The tesla incident highlights the exact technical/ethical problems of using such systems. I was just at a conference speaking with the Google automated car team and the handoff problem is yet to be solved conceptually. The tesla accident is just a first example of it. Obviously the utility of an autopilot system becomes questionable if you can't guarantee safety. It's very similar to medical advances in this way. 1% errors is not a good error rate if it leads to death or serious injury even though such error rates for autonomous systems are remarkable.

That said , if we look at the statistical likelihood of death in a car autonomous systems may very well be much better than humans, but whether we as a society are ethically okay with coming to terms with a piece of software that kills 1 / 100M is another question. A drunk driver that kills someone or themselves is obviously to blame. In this case who's fault is it? The drivers , tesla or the safety board for allowing autopilots to be sold?

I do not want to seem like the guy holding the doomsday banner. I totally agree with you that there are still ways of righting the ship and isolationism is not a contract for war, I am simply pointing out that the similarity to what we have seen in history before is uncanny and we must heed those warnings.

To avoid sounding cliche, I hope we remember history...

pre-WWI / WWII had its fair share of military alliances. None of those alliances were threatened and we know how that turned out.

Imho I think the threat of greater force does not prevent nations from taking drastic actions when pushed against a wall.

Civil unrest is caused by large unemployment especially of the young, which in turn leads to nationalistic and isolationist policies breeding bigotry and violence that are seized upon by demagogues. These are warning signs of something to come. The weird thing is they seem to be popping up all over both in eastern and western europe over the course of the last decade.

The US election in turn I think will definitely tilt the balance. From what i have seen it is not looking good on this side of the pond either.

Cool article.

Something to be said about russia that was not mentioned. The education system of russia has greatly deteriorated. What made soviet science what it was, was the unbelievable school system, from pre school to graduate study. Coupled to a cultural identity linked to knowledge. In many ways, the Russia that emerged out of the soviet union has lost its way in the aspects of learning, compared both to the communist years as well as those dating back to the empire.

This is precisely the case. The integrity of the entire block is now in question as any trade with the U.K. Will now jeopardize some interests internal to the eu. Nationalism is inevitable. Let's hope this does not culminate in war.

I grew up in San Francisco and experienced both public and private school mathematics . The former of which was astoundingly poorly run. It seemed the only way to actually make headway for students that were ahead was for them to study on their own.

The primary issue imo is that math teachers do not seek to inspire students to see mathematics for what is is- a structure way to reason as opposed to a calculation tool replaced by computers. These breeds not only an aversion to the subject by fails to fulfill the basic reason for teaching math.

My best math teacher in school gave us a lecture on chaos - this was algebra class and we were able to generate a fractal by a simple set of coin tosses. This evolved into the discussion of how prbability, physics/randomness of coin tosses and the general notions of chaos and dynamical systems. But such initiative and teachers were never rewarded.

It's sad to see that the education problems seem to be regressing

Very interesting paper. Thank you for sharing it.

I had a question after reading it. The parameter rho, that is used to either classify/regress depending on the t cascade stage{1..T}, is said to be set empirically. This, I would assume, could change across data sets, how were you able to decide when classification switched to regression in your tree and how was this adapted when testing on other data?

Also forgot to attach the paper I was referring to: http://arxiv.org/pdf/1502.02766v3.pdf

I wonder if this points to the vagal nerve as one of the causes for tinnitus. Does anyone know more about this? I looked it up online and just found VNS study completed in 2012 and showed some benefits in a small cohort:

https://www.actiononhearingloss.org.uk/community/blogs/our-g...

Anyone know anything more recent?

It really depends what you are attempting to accomplish, and what you wish to detect in the images.

As you mentioned faces:

Are you looking for face detection or recognition? Face detection has been robustly solved before the advent of DL with HAARs/ face models. Now being pushed a bit further with DL.

(http://docs.opencv.org/master/d7/d8b/tutorial_py_face_detect...)

Current cutting edge face recognition systems rely on DL, and the top performing models are one out of Russia (NTechLAB, facenx_large) and one from Google (FaceNet v8). These were the top two performers in the MegaFace challenge - identification with 1M distractors. Truly remarkable results. http://megaface.cs.washington.edu/results/

As with most DL systems you will need a massive corpus of labeled faces (aka, google or vkontakte - which the NTechLab group used)

Let me preface this that have not used neo in ~1 year.

But before then I used it all the time. Imported the whole US patent DB into it. Its performance was very solid, both for importing and querying using their CYPHER syntax language, which is well tailored for graph representations and graph algos. Definitely had millions of edges, with ~8M nodes.

That said if you are having import issues, you must be not using batch. If the that's the case you are creating significant overheads in your write process (every write acquires certain locks and syncs the db to avoid concurrent mod issues)

I very optimistic about what AR can bring. To me, the biggest roadblock for AR as well with VR, is user interacting with the digital world constrained by the physical one (a problem that's been recognized since the beginning).

Even in AR you do not have haptic feedback, this to me is a critical component for a system to be truly adoptable outside of niche domains such as surgery, industrial processing etc.

I am sure many people are think of this. I have seen some research work out of the Media Lab at MIT trying to address this problem with deformable tables, likewise a Japanese group was doing air pressure haptics.

Could anyone working in this space comment on some state of the art stuff? And provide links? Thanks in advance.

This work looks very cool from the demo video. Have yet to read the paper, but the parametric inversion for generating sounds from features seems very intriguing.

The tundra is primarily always frozen. As you point out freeze/thaw cycles are terrible for biochemical stability, but at the tundra below the surface (where earth acts as an heat insulator) I believe this effect is minimized

DNA has some remarkable stability properties when it comes to preservation. This is one of the reasons people are so interested in DNA storage technologies.

DNA chemistry allows for information to be preserved for millenia at sub-zero centigrade temperatures.

Robert N. Grass, Reinhard Heckel, Michela Puddu, Daniela Paunescu, Wendelin J. Stark. Robust Chemical Preservation of Digital Information on DNA in Silica with Error-Correcting Codes. Angewandte Chemie International Edition, 2015; DOI: 10.1002/anie.201411378

Thank you for the references.

I was not as much interested in mechanism to reduce the training speed by optimizing the structure in order to learn more efficiently. NEAT aims to 'minimize topologies and grow incrementally.'

Rather my interest was in something similar to Q-learning mechanism or RL-LSTM that could learn structural sub-domains within a network, and reimplement them in other portions during training (maybe paralleling NEAT) or during instances of poor output probability.

It is quite interesting that you bring up Hebbian learning, which at its heart is attempting to endow a model with a form of memory stored in the connection or 'edge'.

Current state of the art systems (RNNs, LSTM and ResNets) are taking this one step further attempting to create dynamic memory (much like we believe neural networks work biologically for short term memory storage).

I wonder if the next steps for models are to formulate mechanisms for actually modifying their own graphical structure with a sort of 'protein' based memory, much like the brain is believed to store long term memory? Does anyone have/ or has seen any papers on the topic? Could you provide refs?

I must have missed the post about noisili. This is very cool, thanks for bringing this up.

First off preceptrons are very simplified models, a first pass really at modeling brains in silico decades ago.

There is a whole field of theoretical neuroscience dedicated to this problem, and have made quite a bit of progress, still a bunch of work left to be done.

The classical introductory text on the topic is the book by Peter Dayan and Larry Abbott called computational and mathematical modeling of neural systems.

I wonder if any blind coders could comment on how they see programmatic flow in their minds eye? I am curious how this shapes the way you write code as well. As a demonstrative example, i imagine very long / complex functional methods would be difficult to worth with as they would require more movement back of the screen reader and may reduce your efficiency.

Maybe people who work with blind devs could also chime in.

Many blind musicians and artists develop heightened perceptions in their respective art and generate some unique, original works. I wonder if this is true for coders/ mathematicians?

I think an interesting dichotomy exists here and you touch on it.

Studies that seek to replicate/ confirm work are rarely funded. Similarly studies that are too radical from more standard paradigms are also neglected.

It seems that incremental novelty that can logically be inferred tends to be the most attractive to funding agencies/ review committees.

In some ways it seems a bit like a bell curve with the probability of being funded (y axis) vs the radical nature of the proposed research (x axis).

This is an oversimplified view but something that struck me.

I'd assume the they would be thresholds but the statistics would be interesting.

Had similar experience.

Have you quantified the number of people per account that FB is giving a suggested label to a detected face vs a users number of friends? It'd be interesting to see how FBs classifier performs.

In our case, IP issues dictated for us to get off of university grounds. As anything you build can be claimed to belong to the school if developed on its property.

WeWork provided a good / cheap alternative for a small group of ppl.

The delivery guys definitely- for a while now. Haven't seen that many suits on them. I'm interested what do people in the Netherlands think of them?

The FDA is cautious in this area because historically a number of stem cell injection therapies have led to tumor formations in the CNS.

That said they were not autologous mesenchymal stem cells. This study gives hope.

Refs: http://www.nature.com/stemcells/2009/0902/090226/full/stemce...

http://journals.plos.org/plosmedicine/article?id=10.1371/jou...

Old review from early 2000s http://onlinelibrary.wiley.com/doi/10.1002/ana.60000/full