karpierz

68 ms·

451.
by karpierz 5y ago
Overfitting is, by definition, learning a model that doesn't generalize to the distribution of inputs you care about. If your validation set has the same distribution as the inputs you care about, then your definition holds. But that&#
452.
by karpierz 5y ago
I didn't specify that the training set wasn't representative. All this shows is that you don't need parameters anywhere close to the number of test examples to overfit.
453.
by karpierz 5y ago
I didn't specify the training set, just the test set. It's possible that your model actually models an arithmetic series. Or that it simply overfits. The point is that it doesn't require trillions of parameters to overfit to
454.
by karpierz 5y ago
Suppose that you train a neural network to predict the next number in an arithmetic sequence (a, a+b, a+2b, a+3b, a+4b, ...). As input it gets two numbers, the last number and the current number and has to predict the next one. Suppose you
455.
by karpierz 5y ago
NP-hardness by definition is only yes-no decision problems. The NP-hard formulation of Traveling Salesman is "given the weighted graph G and an integer X, is there a Hamiltonian cycle in G with total weight less than X?"
456.
by karpierz 5y ago
Mergesort worst case is O(nlog(n)). If there's a downside to it, it's that you need to allocate memory.
457.
by karpierz 5y ago
It turns out if you shuffle all of the subsidies and promotions out of your cost center, it's pretty easy to show a profit.
458.
by karpierz 5y ago
> For inference, of course you can use ML to add assumed common knowledge information. I think this is easier said than done, and hasn't yet been accomplished in any general sense.
459.
by karpierz 5y ago
Should the President of the United States have involved himself in race and restaurants?
460.
by karpierz 5y ago
I think if you're going to do an analysis of military strength, you have to do it in context of the political system that produces it. If your society results in supersoldiers, but also causes you to regularly fail at diplomacy and end
461.
by karpierz 5y ago
The investment GP is talking about is money.
462.
by karpierz 5y ago
It's not swapping words, it's swapping message ordering. To swap words, each word would need to be sent in a separate message.
463.
by karpierz 5y ago
> The current winners in these social network markets have basically aligned with one political side. This isn’t even a question any more. In terms of funding of political parties, this is pretty far from the truth.
464.
by karpierz 5y ago
GP isn't saying that he is lucky each game, he's saying he was lucky to be born with the body he has and he was lucky to get the coaching opportunities that came with that.
465.
by karpierz 5y ago
Your efficiency gains will just get eaten up by Jevon's Paradox[1] unless accompanied by a political solution. 1) https://en.wikipedia.org/wiki/Jevons_paradox
466.
by karpierz 5y ago
I don't know that it's really indicative of grace if you phone someone and ask "hey, I'm including quotes from prominent, successful people in a book, do you mind if I add you to it?" and they say yes. You haven&#x2
467.
by karpierz 5y ago
What's the difference between malice and negligence for profit? In the end, the company has deliberately chosen a path which makes them money while creating externalities.
468.
by karpierz 5y ago
So if your brain instinctive recognizes someone as a "her", they're a "her"?
469.
by karpierz 5y ago
How do you tell if someone is a 'him' or 'her'? What are the criteria you use to make that decision, and why are they immutable? (Please don't say XY chromosomes, you don't do a genetic test on everyone you mee
470.
by karpierz 5y ago
On the other hand, there isn't any controversy as to whether blindness is a real condition.
471.
by karpierz 5y ago
I don't think it's an issue of people being interested in how they label themselves so much as society being interested in labels for people. If you come out as gay, suddenly there are a load of assumptions placed on you about how
472.
by karpierz 5y ago
I'm not sure what you're contradicting? In order to get 1 calorie of meat, you need to use 7-10 calories of plants. So by necessity, if you eat 1 calorie of meat, you are producing (at least) 7-10x the suffering relative to eating
473.
by karpierz 5y ago
No, it's not comparing the suffering of plants and animals. It's only pointing out that by virtue of feeding an animal, you necessarily kill plants. So when you raise an animal to eat, you're killing both the animal as well a
474.
by karpierz 5y ago
No, the point is that the process of generating meat requires the animal eat plants. So even if plants suffer, eating animals is still worse, since the animal suffers along with 7-10x the number of plants.
475.
by karpierz 5y ago
Equity is a rough approximation of royalties. It doesn't exactly capture the value of your particular code, but it's pretty hard to figure out exactly what portion of the companies income is due to your code specifically, assuming
476.
by karpierz 5y ago
That's fair, skill is the wrong word. My point was that there are underlying fundamentals (as a stock is partial legal ownership), but they can often be obscured by the noise in the short run.
477.
by karpierz 5y ago
This feels similar to when people describe poker as a game of luck. It's easy to dismiss skill if you haven't played enough hands.
478.
by karpierz 5y ago
You end up with N lists to merge, so to find the smallest head at each given time, you'd take log(N) steps. Given that you have N^2 elements, it would take a total of N^2 log(N) time, which is equivalent in time complexity to simply so
479.
by karpierz 5y ago
In my (limited) experience, the majority of engineers in the valley end up working on MacBooks, so Unix would be the lowest common denominator.
480.
by karpierz 5y ago
Why didn't companies only hire black factory workers during the era of Jim Crow?

More ›