Gábor Melis (@[email protected])

Research Scientist at Google Deepmind #lisp #machinelearning

[Gábor Melis (@[email protected]) @GaborMelis]

@nexta_tv He's a knee and hip expert. After seeing this, I trust he know his stuff.

[Gábor Melis (@[email protected]) @GaborMelis]

Finally uploaded "Adaptive Hashing: Faster Hash Functions with Fewer Collisions" to arXiv for better visibility:

https://t.co/Jm7O8M1Umb

I’ve polished the presentation, fixed typos, and moved to a more readable format from ACM. #HashTables #DataStructures #CommonLisp

[Gábor Melis (@[email protected]) @GaborMelis]

New blog post: Untangling Literate Programming (https://t.co/jog6qK4245):
- The ideal of LP vs its classical implementation.
- How relaxed definition order challenges classical LP.
- Untangled LP, where docs are in the code.
- How Lisp macros and MGL-PAX tilt the scales.

[Gábor Melis (@[email protected]) @GaborMelis]

Peter Medawar, Induction and Intuition in Scientific Thought

[Gábor Melis (@[email protected]) @GaborMelis]

"Most scientists receive no tuition in scientific method, but those who have perform no better than those who have not. Of what other branch of learning can it be said that it gives its proficients no advantage; that it need not be taught or, if taught, need not be learned?"

[Gábor Melis (@[email protected]) @GaborMelis]

Adaptive hashing blog post: https://t.co/dUUFHkXZlW about how to get faster and more robust hash tables.

This is to accompany the previously published materials (see https://t.co/pxNu2Kya9m).

[Gábor Melis (@[email protected]) @GaborMelis]

@newscientist Suppose we don't understand asymptotic arguments. Then, ...

[Gábor Melis (@[email protected]) @GaborMelis]

@RogerGrosse IIRC, funding later was tied to standardized benchmarks. There was lot of apparent progress, which correlated well real-life performance.

[Gábor Melis (@[email protected]) @GaborMelis]

@RogerGrosse Whither speech recognition? by J.R. Pierce, 1969, gives an account of speculative early work in a lush funding environment ("glamour and deceit").

[Gábor Melis (@[email protected]) @GaborMelis]

Why, they are all related to leasehold and freehold, of course.

[Gábor Melis (@[email protected]) @GaborMelis]

New UK citizenship test question: what do marriage value, peppercorns, and the Civil Aviation Act have in common?

[Gábor Melis (@[email protected]) @GaborMelis]

@rlmcelreath TLDR, but the excerpt seems to want to express that posteriors shall contract with more evidence, and physicists have seen much. A good armchair and you're set for life.

[Gábor Melis (@[email protected]) @GaborMelis]

But d-separation hence conditional independence is much more conspicuous on the plate ... unless there is just too much on it.

[Gábor Melis (@[email protected]) @GaborMelis]

Plate notation lacks crucial details and has limited expressiveness. In https://t.co/qLLy9tuT3H, Rob Zinkov argues, and I agree, that stating the generative model like this in text much better:

[Gábor Melis (@[email protected]) @GaborMelis]

@thx3188 @Kawa_oneechan @Foone And of course, if you're married, you'd also have to file the K-7209 Statement of Joint Intentions and declare any mutual gift expenditures.

[Gábor Melis (@[email protected]) @GaborMelis]

@thx3188 @Kawa_oneechan @Foone You have until midnight on the twelfth working day past the first full moon after the end of your fiscal year. However, you can extend the date by filing an RPM-78 Waning Interest Extension anytime before the close of business on the second Tuesday after the first Friday in March

[Gábor Melis (@[email protected]) @GaborMelis]

@leopoldasch Believing in this straight line also requires ignoring the second derivative. Also, further tweaking of the Y axis should make it even straighter.

[Gábor Melis (@[email protected]) @GaborMelis]

Absolute value

[Gábor Melis (@[email protected]) @GaborMelis]

Slides for my talk on Adaptive Hashing: https://t.co/GwJ7zVKa9v. Also, the LaTeX sources of both the slides and the paper are available at https://t.co/pxNu2KyHYU.

[Ben Recht @beenwrekt]

I wrote today abot statistical rigidity and losing the plot. https://t.co/u9vokobHKB