nightcracker

13 ms·

61.
by nightcracker 5y ago
A counterpoint to this is that if you wish to recreate many of the classic (and modern) sounds, that often they were created with subtractive synthesis, the easiest path to a close sound is to also use subtractive synthesis. If you wish to
62.
by nightcracker 5y ago
> Agreed on pdqsort being 3x slower worst case on "killer" input, but it's my understanding that's why it's not being used to replace std::sort. This is just not true. Introsort (the current std::sort in all C++
63.
by nightcracker 5y ago
What do you mean by "suffers from killer inputs"? It is always O(n log n) by falling back to heapsort, so I don't think any input can be described as a "killer".
64.
by nightcracker 5y ago
> This is where the skew (can) come from. No it can not! It's just mathematically and biologically not true, and attempts to erase a crime against humanity. The human inception process does not give a shit about your decision whethe
65.
by nightcracker 5y ago
You simulated the limiting case, where you don't stop having babies until you get a boy. That's (obviously) not realistic as we have a finite lifespan and fertility, but note that the expectation still holds. In this process you w
66.
by nightcracker 5y ago
No it's not. As far as we know the chance of a child being male or female is 50% (not quite, but I'll assume it for simplicity here), independent from any previously born children. That is, it is a 'memoryless' process,
67.
by nightcracker 5y ago
Time to include a hidden 10000x10000 white PNG to all your webpages I guess.
68.
by nightcracker 5y ago
When working with real-world statistics? Absolutely, there is no reason to believe that this assumption holds. But the context is randomized (quantum) algorithms, where this is trivial to guarantee.
69.
by nightcracker 5y ago
Fun fact that I had to prove for my master's thesis: if you have a procedure that (independent from other estimates) estimates a mean with absolute or relative error eps with probability 1/2 + g, then you can boost that to an arbi
70.
by nightcracker 5y ago
Most everything with a real number answer in quantum computing is "probably approximately correct". That is, with probability delta the answer is no more than epsilon away from the correct one.
71.
by nightcracker 5y ago
The problem is that your "See how it works" link prompts people to log in. I don't want to log in, so I guess I'll never find out how it works, or download the app.
72.
by nightcracker 5y ago
The authors do claim to do a hyperparameter sweep but only for vanilla XGB hyperparams.
73.
by nightcracker 5y ago
Batch normalization is nothing neural network specific to it if you use it on the input layer. I don't think it matters for a tree algorithm like XGBoost either way though. SWA is pretty NN specific. So leave it out for XGB. There&#x27
74.
by nightcracker 5y ago
> They test against XGBoost, GBDT Auto-sklearn, and others. Did you read the paper? Yes. Did you read my comment? They compare NN + Cocktail vs. vanilla XGB. They don't compare NN + Cocktail vs. XGB + Cocktail. To make it crystal cl
75.
by nightcracker 5y ago
Paper is missing the control: how good is this 'cocktail of regularization' when applied to traditional methods like XGBoost? At best you can claim the result here that neural networks with regularization methods can beat traditio
76.
by nightcracker 5y ago
And how is everyone supposed to know when to leave their room and where to go?
77.
by nightcracker 5y ago
I took a course on this and similar techniques called "Information Theoretic Data Mining". You can see a bunch of relevant references on the course site: https://eda.liacs.nl/teaching/itdm/ .
78.
by nightcracker 5y ago
It was cheap only in the "penny wise" way.
79.
by nightcracker 5y ago
At least when using Gaussian white noise, the DFT of the noise is the same distribution with a smaller variance: https://dsp.stackexchange.com/questions/24170/what-are-the-s... I don't think the same neat res
80.
by nightcracker 5y ago
Game textures often use this kind of noise for terrain heights, smoke, etc. A similar kind of noise known as blue noise can be generated by taking the Fourier transform and not applying a 1/f filter but a high-pass filter instead. You
81.
by nightcracker 5y ago
With the advent of online-only games what we'll eventually see is the erasure of a generation's (sub)culture. When in 2070 a kid will ask their parents what they used to do when they were small, they'll answer "Play Leag
82.
by nightcracker 5y ago
You can construct arbitrarily-sized permutations with log(n) time random access using Sometimes-Recurse Shuffle: https://eprint.iacr.org/2013/560.pdf .
83.
by nightcracker 5y ago
I'm the author of that code. I have a semi-detailed writeup here: https://drive.google.com/file/d/17aK9ifP79abRuafXSD24jVVI3Sb... Within one Fenwick tree we have two sorts of paths starting at any node, one g
84.
by nightcracker 5y ago
Only if you hold the religious belief that capitalism and its consequences are inherently beyond critique.
85.
by nightcracker 5y ago
If you know a product is faulty, why did you choose to buy another one?
86.
by nightcracker 5y ago
Damnit, after the mistake of calling the study of computation 'computer science', I had thought we'd avoided the issue with 'quantum computing'. But no, even this term seems to get muddied by things that aren't
87.
by nightcracker 6y ago
And if you have a datacenter with 200 machines, that's roughly once a month.
88.
by nightcracker 6y ago
Beyond that, tmux is (for me and probably many others) critical software used constantly when SSH'ing into many different boxes. Different boxes running different versions of tmux. Making it impossible to share one config file everywhe
89.
by nightcracker 6y ago
You have to be one kind of stupid to casually travel with that much cash and a whole 'nother kind of stupid to tell the person you're sitting next to that you're doing so.
90.
by nightcracker 6y ago
Have you ever used Python? If you did you really wouldn't be saying this. There isn't an exception. The semicolon is used to put multiple statements on a single line. That's it's only use, and that's the only time i

More ›