HNHacker News
TopNewBestAskShowJobs

discardorama

3,407 karma · joined July 3, 2013

submissionscomments
discardorama··on WikiLeaks Begins Releasing Leaked Saudi Arabia Cables
> this doesn't say anywhere that the Saudi government supports ISIS.... if anybody knows two shits about the middle-east, he/she wouldn't say something like that.

KSA is a totalitarian regime. Do you think the people would be able to fund ISIS against the wishes of the government? This is a government that doesn't even let women drive and lashes bloggers for mild criticism. And yet you claim they didn't know/approve of people funding the ISIS? I'm afraid it's you who a little ignorant about the middle east.

The Saudis may not fund the ISIS now; but they did heavily fund it it and helped set it up. The aim was to destabilize and topple Assad. Just like they did with OBL and Al Qaida; and, predictably, got similar results.

discardorama··on WikiLeaks Begins Releasing Leaked Saudi Arabia Cables
Saudi Arabia is at the center of a lot of stuff going on in the middle east right now (their support for ISIS, their involvement in Yemen, the chess game with Iran, etc. etc.). It would be interesting to see what these cables turn up, once translated.
discardorama··on WikiLeaks Begins Releasing Leaked Saudi Arabia Cables
I assume the implication is that Wikileaks is somehow in cahoots with Russia?
discardorama··on Inceptionism: Going Deeper into Neural Networks
My understanding: when you're doing standard gradient descent, you push the error down through the layers, modifying the weights at each layer. Now, in "normal" NN training you stop at the input layer; it makes no sense to tweak the error at the input layer, right?

But what if you did the following: flow the error down from the outputs to the layer you're interested in, but don't modify the weights of any of the layers above it; just modify the values of this layer in accordance with the error gradient.

Added later: I think we should wait till @akarpathy comes along and ELI5's it to us.

discardorama··on Ask HN: What are the greatest discoveries in the last few years?
> Another example is nuclear energy. Clean, abundant, and extremely safe.

Yeah, but what do you do about the waste? The US has tons of waste sitting around. The grand plan of using Yucca Mountain to store the waste fizzled. We have just postponed the cost to a future generation who'll have to deal with the waste we're producing today.

discardorama··on An Insider’s Guide to Shenzhen Manufacturing
Someone did a global search-and-replace of "make" with "Make" ...?
discardorama··on How Harry's Bought a German Razor Factory
I tried Harry's when they first started. I was excited at the prospect of ditching Gillette. For some reason, I get nicks and cuts when I shave with a safety razor (the last time I tried). I do shave after showering, to keep the hair soft.

But Harry's just didn't cut it (no pun intended). I got many more nicks and cuts with Harry's than I did with Gillette. Plus, their shaving cream gummed up the razor, and cleaning was a pain. So after a few days, I had to go back to Gillette, and Harry's is sitting in the corner somewhere. Maybe it's their 5-blade monster; I don't know.

discardorama··on What's Wrong with Deep Learning?
Newborn babies can hardly do anything because their parameters have not been set; but their hyperparameters (things like, to use NN analogies: activation function, learning rate, etc.) are baked in.
discardorama··on What's Wrong with Deep Learning?
> Max pooling tests if a feature occurs anywhere in a certain area, rather than being sensitive to the exact location.

From Geoff Hinton's AMA on Reddit: The pooling operation used in convolutional neural networks is a big mistake and the fact that it works so well is a disaster.

discardorama··on What's Wrong with Deep Learning?
But not in the same brain. Sure, evolution happened and picked the right parameters; but today's brain comes with the hyperparameters baked in (with some small amount of randomness). It is able to do all the things it can do without the luxury of parallel training and tuning.

For example: you don't need to show a baby a 1000 photos of mugs before it can tell what's a mug and what isn't. Just show it 1 example of a mug a couple of times, and from then on it's able to identify mugs and mug-like objects pretty reliably.

discardorama··on What's Wrong with Deep Learning?
If someone were to ask me what's wrong with DL (and not that anyone would, since I'm an unknown), I'd say the lack of theory. Most DL results look very hacky to me. Someone says Max Pooling works; someone again comes along and says it's not necessary. Someone says sigmoid or tanh are the best activation functions; someone else says ReLUs are better. And so on. Why? Why is one better than the other?

I'm no biologist, but I don't think our brains are going around trying to do a grid search for the best hyperparameters. Most DL results today are the result of throwing 1000s of Titans on the problem and then sitting back for a week for the beast to cough up a solution.

Tangential nitpick: one (very minor) nit I have with Prof LeCun's presentations is that I don't see him give more credit to Hinton and Schmidhuber. Hinton is mentioned a couple of times (3), but Schmidhuber is totally ignored; for example, when he mentions LSTM, it's cited as [Hochreiter 1997], even though it was a join publication with Schmidhuber. It should be cited as [Hochreiter et 1997], as he does in the very next line.

discardorama··on What's Wrong with Deep Learning?
As far as I remember, everyone was jumping on top of SVMs and Markov Models.
discardorama··on Why the Internet in Sub-Saharan Africa Sucks
Summary: Nairobi needs a datacenter, with Akamai and Cloudflare.
discardorama··on Men in committed, romantic relationships have lower testosterone
(Sample of 1). I think my personal experience would seem to indicate so. I've been in a committed relationship for 2 years. But sometimes the wife travels, and I hang out with my single friends, going to bars, flirting with chicks, etc. And I find that I'm a lot more 'affectionate' after this episode. Could it be that I miss the wife so much? I have noticed a difference between when I do the bar scene, and when I don't.
discardorama··on Why Go is doomed to succeed
I don't think you can say "Reminds me of Java" with "Simple enough". Java is too complicated now.
discardorama··on Facebook Now Cares About How Long You Look at Stuff in Your News Feed
"Dwell time" (as it is known in the industry) has long been a feature at major search engines and news portals. It is a pretty good signal of relevance.
discardorama··on The Greatest Bullfight Ever (1948)
I don't know how a culture like Spain's can call this a "sport". The whole game is rigged against the bull; if the bull does get an upper hand, he is quickly shooed away and the matador saved. And I also can't understand how the matadors can act all macho and stuff, when they have the whole cavalry behind them, so to speak.

It'd be closer to being a 'sport' if it was just the matador against the bull, mano a mano.

I'm very glad that Catalonia has banned bullfighting: http://www.nytimes.com/2011/09/24/world/europe/adeu-to-catal...

discardorama··on A venture capital firm says it invests in people “pre-idea, pre-team”
Using a throwaway account to attack someone is not cool. If you have the balls (or the equivalent for the other gender), stand up and make the comment in your own name.
discardorama··on Inverting Binary Trees Considered Harmful
Apparently it means to swap the left and right subtrees. How's that "inversion"? I'd call it "flipping" or "mirroring" or something... If this is the case, then it's a trivial solution: invert(left_child); invert(right_child);
discardorama··on Inverting Binary Trees Considered Harmful
> I'll be the contrarian and claim that inverting a binary tree is a perfectly reasonable interview question.

The fact that so many people in the previous discussion could not even tell what that means is a very bad sign.

discardorama··on Inverting Binary Trees Considered Harmful
> The recruiter tells you months before the interview and they (Google) even send you a list of what books and papers to read before you consider yourself ready.

Right, because I have nothing better do in those months than to rehash CS 101 and waste my time inverting trees and reversing strings.

discardorama··on Google: 90% of our engineers use the software you wrote (Homebrew), but...
> Google isn't just a search company anymore they're working on lots of very interesting non-trivial things all around the company.

Right. And I really doubt that people like Andrew Ng or Geoff Hinton or Sebastian Thrun were asked to invert binary trees....

discardorama··on I quit the tech industry
I was wondering what this person's situation was. Then I read this in her(his?) blog:

> I currently work for Yelp from the comfort of my home in Las Vegas. I live with my ridiculous cat, my partner, her three cats, her husband, his dog, and our boyfriend.

Now it makes more sense. In my limited experience, when you're working remotely, you feel disconnected from the mothership. To this, add some work stress, and it's easy to get disenchanted.

PS: I don't really know how to parse that sentence. No wonder NLP is so hard! ;-D

discardorama··on How MIT Students Won $8M in the Massachusetts Lottery (2012)
As a broke grad student, I did the same; except, I never really bet anything. I would put $20 in the slot machine, and promptly hit the "drink service" button. Once the drink came, I'd hit the "cash out" button, collect my $20 and move on. I know this sounds terrible, but (a) I didn't have the spare cash to throw away at the casino, and (b) I didn't have the skills to actually play any of the games where skills counted (like poker or blackjack).
discardorama··on How MIT Students Won $8M in the Massachusetts Lottery (2012)
But a casino does not consider counting cards legal. Hence the boot.

Consider, for example, one of those "how many marbles are in this jar?" games. If I use my knowledge of math and physics to derive a pretty accurate estimate, it's not wrong, is it?

discardorama··on How MIT Students Won $8M in the Massachusetts Lottery (2012)
The only time I've "won" anything in a casino is when I hit the "drink service" button to get a free drink...
discardorama··on How MIT Students Won $8M in the Massachusetts Lottery (2012)
> The problem is the same as that for insider trading on the stock market:

Nope, not at all. These people are playing by the rules. The lottery officials are aware of their play; and yet don't consider it illegal. There is nothing wrong with using your superior skillset legally to make some money.

discardorama··on Holding Your Breath in India
My point was that even in the US, fixing an environmental disaster (and a small one at that) took a while. It didn't happen overnight.
discardorama··on Holding Your Breath in India
You're blaming the messenger here.

I travel to India a couple of times a year, for 2-3 weeks at a time. As the plane is landing in DEL, you can smell the difference. As soon as you step on the jet bridge, you get acrid smell of burnt ... something. Rubber? Plastic? It's hard to tell (and you're overwhelmed by the smell of phenyl used for cleaning).

Every time I've taken a 2-wheeler or a 3-wheeler in the busy streets, I come home with a cough; and near the end of the trip, I have a consistent burning sensation in the throat.

discardorama··on Holding Your Breath in India
Agreed. I'd get the hell outta dodge at the first sign of trouble for my child. Steroid therapy -vs- clean air?!?! No comparison: clean air wins every time!
← PreviousPage 11 of 29Next →