brilee

25 ms·

241.
by brilee 9y ago
News at 11: When you kill bacteria via technique X, some bacteria are affected by it more than others. Since not all bacteria are killed, this leads to the enrichment of some types of bacteria. When a scientist took the most populous bacter
242.
by brilee 9y ago
From the article: Unlike floats, ints are represented as immediates, i.e., they are not allocated on the heap. To tag ints OCaml steals the LSB bit of an int and sets it to 1 always. This means that we get 63 bits of precision in OCaml ints
243.
My Path to Machine Learning(moderndescartes.com)
2 points by brilee 10y ago|0 comments
244.
by brilee 10y ago
Video of the AI here, playing as the black captain falcon: https://www.youtube.com/watch?v=dXJUlqBsZtE
245.
by brilee 10y ago
My take on it is, even if health costs were 30% of my cushy software salary, I'd pay them, because well, everyone wants to live. And so, prices continue rising to the point where people are willing to pay for them. I'm willing to
246.
by brilee 10y ago
To give an update on status of project, I'm currently at the point where the policy net alone plays at 2-3kyu with a day or so of training. MCTS is implemented but Python is slow enough that I don't get significant Playouts. As a
247.
by brilee 10y ago
Not quite; I would describe it as moving the complexity into the algorithm. For example, even though the bitpacking from the post achieves fast good compression, it is unable to compress floats in general; merely 1.0 and 0.0. The knowledge
248.
by brilee 10y ago
It amazes me that Haskell takes my Python hack and actually allows it to be expressed quite naturally! I'm curious what the semantics are for the interleaving of do and guard - would you say that Haskell will compile that code to the e
249.
by brilee 10y ago
Author here. If I may, could I ask you to write the equivalent code as you have suggested? I can see one way to implement with generators but it's not particularly concise. The benefit of the list comprehension approach is that its mec
250.
by brilee 10y ago
I had an opportunity to dig into the details of hyperloglog recently, and I wrote up my results here - http://moderndescartes.com/essays/hyperloglog HyperLogLog is definitely one of the most elegant algorithms I'm
251.
by brilee 10y ago
The algorithm was a clustering algorithm, and takes as input the number of clusters desired. The researchers put in 50 clusters, so at some point the algorithm declines to separate sparse regions, because the separations are better used els
252.
by brilee 10y ago
This video has nothing to do with the title "without a PhD". It's a walkthrough of basic techniques for training neural networks on the MNIST handwriting dataset.
253.
by brilee 10y ago
SMILES strings are not necessarily unique; each molecule can be encoded multiple ways as a SMILES string. As a sanity check, you could pass multiple representations of a molecule into the autoencoder to see if they generate the same continu
254.
by brilee 10y ago
The last two paragraphs of the article says it all - the FDA has embargo power over the media, because the media can't cooperate to set their own embargo: Even a close-hold embargo wouldn't constrain a reporter without the rep
255.
by brilee 10y ago
I recently replaced my paid SSL certs using LE. I took what I would have paid in SSL costs over the next two years and donated it instead.
256.
How (Not?) to Use Python's List Comprehensions(moderndescartes.com)
2 points by brilee 10y ago|0 comments
257.
by brilee 10y ago
Neat! It should be noted that the genetic algorithm used is not the same as reinforcement learning, in which the entire history of decisions leading up to a success is fed back to the NN as training examples.
258.
by brilee 10y ago
There's this structured data format called XML, and typically you put a XML schema definition at the top (example: <xs:schema xmlns:xs=" http://maven.apache.org/xsd/settings-1.0.0.xsd"> for a Maven
259.
by brilee 11y ago
silicon is diamond cubic; 0.54 nm corresponds to 8/sqrt(3) radiuses worth. So the diameter of silicon is 0.23nm, and 7nm = ~30 silicon atoms across.
260.
by brilee 11y ago
Not unless the fuel contains only carbon - as charcoal does.
261.
by brilee 13y ago
Question: I see all this discussion about buffering and encoding and such. In python, this seems to work: import sys sys.stdout.write(sys.stdin.read()) What is being abstracted away/handled by python that I'm not seeing?
262.
by brilee 13y ago
I wrote a similar game called Fibs, where the game mechanism is that adjacent Fibonacci numbers are combined to make the next Fibonacci number. It's also command-line based. https://github.com/brilee/games/blo
263.
Fibs (Yet another Threes knockoff)(github.com)
1 points by brilee 13y ago|0 comments
264.
by brilee 13y ago
Another popular method of doing this is a variant on the area mapping technique, except that instead of taking the weighted average of the four nearest pixels, it instead creates a 2-D spline across the R/G/B "heights" o
265.
by brilee 13y ago
How is this different from an escrow service?..
266.
by brilee 13y ago
In a land of immortality, would you willingly give yourself an 80-year death contract in order to motivate yourself to live your life?
267.
by brilee 13y ago
If you want to help people organize and get together without being underhanded, you're more than welcome to set up your own organization.
268.
by brilee 13y ago
This is the first time I've heard that iMessage and Facetime are end-to-end encrypted. Can somebody comment on that? I imagine it only works if both ends of the conversation are on iMessage.
269.
by brilee 13y ago
Of course ancient Roman concrete is better than what we have now. The ancient Roman concrete that wasn't better than what we have now, didn't survive to today. There was probably a wide range of random kinds of concrete, and a 200
270.
by brilee 13y ago
http://askakorean.blogspot.com/2008/09/why-do-koreans-compla... http://askakorean.blogspot.com/2012/12/quick-thought-about-p... Yes, the US are in fact occupying South Korea, and South K

More ›