olooney

35 ms·

271.
by olooney 7y ago
I did Strang's linear algebra course[0] (link goes to Youtube playlist of lectures) several years after graduating and recommend it highly. I was looking for refresher but I gained a deeper understanding of several important concepts;
272.
by olooney 8y ago
Not knocking on this author at all... It's just that nowadays if I see a title in the vein of "7 Tips to Train Deep Neural Nets for Complete Beginners On Rails With Keras and TensorFlow" I click on it more out of a sense of o
273.
by olooney 8y ago
Well, this was unexpectedly excellent. I don't think "stick with supervised learning" is very good advice, though. Unsupervised techniques sometimes work well for NLP and has worked well for other domains, such as medical rec
274.
by olooney 8y ago
I earned a masters in physics about 15 years ago and started in industry as a software developer. Over time I started working on more analytic applications, basically because I could understand the math, and for the last six years I've
275.
by olooney 8y ago
Don't forget the Gompertz function[1], used to model population growth: $$ g(x) = a e^{-b e^{-cx} } $$ [1]: https://en.wikipedia.org/wiki/Gompertz_function
276.
Where Electronic Health Records Went Wrong(fortune.com)
3 points by olooney 8y ago|0 comments
277.
by olooney 8y ago
Ironically, it's much easier to turn gold into lead. Simply leave a bar of gold near a fission reactor or other neutron source for a couple of years, and eventually it will turn to lead.
278.
A Leprechaun Explains the Usefulness of Non-Existent Mathematical Objects(rjlipton.wordpress.com)
3 points by olooney 8y ago|0 comments
279.
by olooney 8y ago
Very interested article. I've often railed against putting your thumb on the scale (or even worse, second-guessing) machine learning models by applying too many so-called "business rules," especially post hoc rules. If the
280.
A Fairly Fast Fibonacci Function(oranlooney.com)
131 points by olooney 8y ago|58 comments
281.
by olooney 8y ago
Proof that at least one number exists: Take for our logic first order logic with equality[1]. Take for our definition of the word "number" the Peano axioms[2]. The first Peano axiom is "0 is a natural number." Define &qu
282.
by olooney 8y ago
I would guess both you and author of the article have in mind something like gene expression[GE] in bioinformatics. Thousands of markers, but only hundreds of examples, and the researchers are using some automated feature selection approach
283.
by olooney 8y ago
> decline of formalism as a credible philosophy of mathematics I suppose someone who's writes a paper about mathematics anticipating physics would not be a formalist, but to say formalism is not credible is a bridge too far. As far
284.
by olooney 8y ago
Thank you, good clarification, sorry if I made it sound I was dog-piling on Ng. Andrew Ng works harder at explaining ML in a non-technical way than anyone else I can think of, and the 1-second rule came out of that genuine desire to educate
285.
by olooney 8y ago
Not this again. Haskell, Prolog, CSS, SQL are classic examples of declarative programming, but really, what isn't? Everything except machine code is to a certain extent declarative. Obviously, even the C compiler is free to optimize an
286.
by olooney 8y ago
Andrew Ng once tweeted about a 1-second heuristic: > Pretty much anything that a normal person can do in <1 sec, we can now automate with AI. https://twitter.com/andrewyng/status/788548053745569792?lang... A
287.
by olooney 8y ago
> breakthrough innovation in programming which will enable programmers to deliver ambitious web/mobile apps at a very high speed. That's a pretty narrow framing. 10 years from now, a lot of programming won't look like trad
288.
by olooney 8y ago
Some very nice ones in here: AVL and Red/Black trees, especially the rotation step. I also liked it when the code shows up on the upper left and you can see it stepping through line-by-line, such as the Fibonacci example. The radix tre
289.
by olooney 8y ago
Apocalyptic? Science would first grind to a halt, reverse itself, then descend into a thousand-year dark age of sophistry, apologetics, and superstition. Alexander Fleming and Jonas Salk would be laughingstocks, Vaccinations would be banned
290.
by olooney 8y ago
UMAP[1] and t-SNE[2] are two very similar and popular ways to map a set of points in a high dimensional space to a low-dimensional (usually 2D or 3D) space by assuming all the points are close to some smooth curved surface (a manifold[3]).
291.
by olooney 8y ago
It's always insidiously attractive to deny voting rights to groups of people who, in our own opinion, do not have good character or good judgement. It certainly sounds clever to say we're implementing rules so only the best and
292.
Reverse-mode automatic differentiation(rufflewind.com)
3 points by olooney 8y ago|0 comments
293.
by olooney 8y ago
It's incredibly common. For example: https://docs.google.com/spreadsheets/u/1/d/e/2PACX-1vRPiprOa...
294.
by olooney 8y ago
This claim is unsupported: > The machine, not smart enough to do the actual difficult job of converting these sophisticated image types to each other Obviously it found an easy way to solve the problem it was given: stenography But coul
295.
by olooney 8y ago
I assume that the lone developer locked in the garage is still able to communicate as much as necessary with clients and stakeholders. The garage metaphor makes it sound like they're cut off the whole world, when what I intended was ju
296.
by olooney 8y ago
People who know more about deep learning than I do tend to argue that there is empirical evidence that non-convexity is a non-issue because the performance of local minima will be close to the performance of the global minimum, given a suff
297.
by olooney 8y ago
Here is a (tongue-in-cheek) model I've developed. It's based on the observation that teams who were basically competent and had an overabundance of hard technical skills reliably delivered, while teams where hard skills were rathe
298.
by olooney 8y ago
For the transition from Python 2 to 3 the Python team released just such a conversion tool, aptly named "2to3." [1] This was better than handy but the transition from 2 to 3 was still lengthy and painful. In practice many of the s
299.
by olooney 8y ago
If you want to learn more you can Google the top vendors in the DMP space: Krux (now Salesforce DMP), BlueKai, LiveRamp, DataLogix, Acxiom [sic.] A couple of these offer "CRM Onboarding" (matching real names/addresses to anon
300.
by olooney 8y ago
I saw Persi Diaconis lecture on coin flips and shuffling cards once. Very interesting guy. What metaphor for a Bernoulli trial do you prefer?

More ›