HNHacker News
TopNewBestAskShowJobs

ilzmastr

155 karma · joined April 2, 2013

submissionscomments
ilzmastr··on Show HN: Neural network color recognition
Good stuff.

quick q: why do you call this a hopfield network? I see you have a fully connected 2 layer neural network (0 hidden layers): https://github.com/mateogianolio/hopfield-color-recognition/...

instead of a bunch of circularly connected perceptrons: http://en.wikipedia.org/wiki/Hopfield_network#mediaviewer/Fi...

p.s. it looks like the library you're using has a built in Hopfield network `new Architect.Hopfield(10)` (at the bottom: [here](https://www.npmjs.com/package/synaptic)), why didn't you use this?

ilzmastr··on Show HN: Neural network color recognition
Interesting stuff, nice work!

The complex pattern is a result of the network doing the equivalent of fitting 3 gaussians (1 for each of RGB color channel) on the intensity of the pixel in that channel. I was curious, so I recorded a tiny demo of doing the equivalent piecewise in photoshop: http://cl.ly/183Z3w1V1B0F

Posterization in photoshop is a k-means technique, and each k-mean is a center of a gaussian, hence my motivation for the analogue video.

The triangles in the thresholded image make me think that the network is doing the equivalent of predicting a probability that a certain pixel is white/black per channel, and then these channels collude for the final decision, and thus we see triangles when the intersections of the 3 triangly spaced gaussians yield a value above some threshold.

Pic of the different channels: http://cl.ly/image/1w371V3U0N02

ilzmastr··on Hard Drive Data Sets
thanks for sharing this! A few quick questions:

- Do you guys do any precise prediction on if a particular drive will fail soon and replace it?

- I notice a lot of sparsity in some rows, that is different than a 0 in that field I assume? Does that mean anything else interesting?

- Also under the "inconsistent fields" section you say "drive manufacturers don't generally disclose what their specific numbers mean," can you give a hint as to one of the drive models that has a minimally sparse smart readout and has information available from the manufacturer on what those smart numbers signify?

I figure if anyone has collected the references on what the metadata means, and for which models it is available, it's you guys :)

ilzmastr··on The Secret History of Silicon Valley (2010)
how is this not an epub? Here it is, with all images, 26MB: http://cl.ly/3L202v3T3i0J

made w/ nokogiri and pandoc

ilzmastr··on On Eigenfaces: Creating ghost-like images from a set of faces
Classic HW problem nicely done. Today they use Viola-Jones.

For those wondering why PCA works (self-plug): http://ilyakava.tumblr.com/post/95691347612/demystifying-pca

ilzmastr··on Coding is not the new literacy
Besides the title being pedantic, I feel this article is alienated from the average person I know who is always modeling and dealing with symbols and representations in the world, but less often takes a step back to realize the practice of their modeling and to judge its efficacy.

To me, modeling seems equivalent to consciousness. To me cave paintings of animals seem significant because they illustrate 41k yo humans creating representations of their world (for mental practice, entertainment, ceremony, who knows...), whether they realized their practice is another question.

But today, there are many people who model and recognize it, think of manipulative salespeople/lawyers/cads, great fiction writers, constipated philosophers and art/film/lit critics, scientists, artists/architects/designers...

You don't __need__ any technology built after 1980 (or whatever) to explore the world... Even thought experiments can shake an academic field (Maxwell's demon, Schrodinger's cat, EPR paradox...)

I agree that there may be better tools to convince people to contemplate environments and unpack/model/break down confusing things into collections of simple theories than to send them to a liberal arts college (where this and goofing off is really all there is to do in my experience)...

But programming is still the new literacy IMO. I feel like all the white collar underpaid people I know could work 2-4 hours a day of their 10 hour/day jobs if they wrote the most limited, banal, imperative script that abstracts no concepts and in completely un-generalized (and therefore fails at modelling). And I feel just this fact, that there would be a great liberation of 1st world mankind's time if every non-programmer who uses a computer for their job up and wrote a severely shitty program that automated some aspect of their responsibilities, is enough to say that a coding is the new literacy.

Think of what kind of intellectual shift a machine-literate public that just cut their work week in half would result in...

ilzmastr··on David Foster Wallace's Kenyon Commencement Speech (2005)
http://en.wiktionary.org/wiki/university_of_life
ilzmastr··on What are your favorite sci-fi and fantasy books?
Childhoods end - Humanity takes the next step in evolution

Anhilliation - Fish out of water like story of exploring a mysterious, horrific, environment (2 sequels drag on for too much after this one though)

Many Ray Bradbury stories (Martian chronicles, and the Everyman collection contain the best ones)

Infinite Jest (Alternate American future of a culture of addiction and depravity, partly mediated by technology)

The Circle (haven't read, but like the author, life inside the biggest tech company of the future)

ilzmastr··on Gotham Air: Manhattan to JFK in 6 minutes for $99
Slightly cheaper than cab fare+tip from Manhattan to Newark, but ~40 mins faster. Then again you still have to get through financial district to the helipad...

Still not as good as having a private helicopter. I hear NYC to PHL is 30 mins by heli.

ilzmastr··on Show HN: New Hacker News web client
Nice! The live update is sick!

What would make me use this instead of news.ycombinator.com every day however would be:

- the ability to hide stories you've already read (some box to tick like a todo list) or are not interested in. They take up so much space, so there is a lot of wasted space after 5 mins on the site and reading/seeing I'm not interested in the top stories. I don't want to click 'show more', I want new headlines to replace old unwanted headlines

- Color coding the actual divs to signify hotness and using size of divs to signify importance/upvotes would be nice. The font is so thin in the orange circles (and the spacing between them and the text) that I cannot read it in my peripheral vision and have to glance back and forth between titles and ranks

- hover movement is disturbing enough for me to navigate with keyboard instead of mouse

ilzmastr··on Machine Learning Course by Tom Mitchell
I was in a similar situation a year ago. Two things fixed my problems:

I took a course on Linear Algebra (Bretscher's book up to chap 9) and a Probability course (Ross' book up to chap 6) and did very many problems by hand on paper. I just finished a ML course (Bishop's book, and Jordan's book), mixed grad+undergrad, which was 80% problems w/ pen and paper and 20% code up something algorithmically trivial but mathematically challenging, and don't think I would've been able to pick up the additional math along the way without these two great books and their many exercise problems behind me.

I read layman's explanations of ML concepts a year ago, and got nowhere in terms of my own implementation+debugging/improvement upon established techniques. Now I can solve problems I saw a year ago and thought "no one can do this."

My advice is to take the gateway drugs first, Probability (Ross <- I love this book!) and Lin Alg (I like Bretscher much better than Strang, but not everyone agrees with me :) Take a course in real life (for a grade and a transcript) at a competitive university if possible, nothing makes you study thoroughly like a gun to your head.

ilzmastr··on A Tutorial on Support Vector Machines for Pattern Recognition (1998) [pdf]
Yep. Was assigned reading in my ML class this semester. Loved it.
ilzmastr··on Show HN: Hacker News iOS client using the new official API
iOS 8 only! :( Wish I could try it! Screens look great!
ilzmastr··on The Happy Ending Problem [video]
Reminds me of VC dimension kind of stuff for describing binary classifiers.

That first example w/ 5 points necessitating a convex 4 sided polygon is like the VC dimension proof that an axis aligned rectangle can never correctly classify 5 points, since you can never arrange 5 points w/o one of them being in the body of a rectangular region surrounding the other 4 (and that interior point's label therefore can't be arbitrary and still correctly classified). more: https://www.cs.princeton.edu/courses/archive/spring13/cos511...

Wonder if the 2^(n-2) + 1 rule is somewhere in the VC dim lit

ilzmastr··on An Intro to Videogame Design History
made an epub of the 4 parts with images for my own read-it-later use b/c I'm obsessed with pandoc:

http://cl.ly/2I0f0i1a2d3u

ilzmastr··on Sparklines for your shell
Nice! A more traditional version of this in the shell that I love to use day to day is bashplotlib: https://github.com/glamp/bashplotlib
ilzmastr··on Show HN: A mostly complete 2014 "Tools of the Trade"
waffle.io is hyperdope on the Bug/Issue Tracking side too. Makes GH issues look like trello, and allows boards for multiple repos.
ilzmastr··on Every line of code is always documented
Interesting concept. Coincidentally, I wrote a gem recently that serves the same function as the git-overwritten script at the end: https://github.com/ilyakava/stefon

gem install stefon

ilzmastr··on Isolating spurious and nondeterministic tests
Good way to motivate better test-writing. A leaderboard of the authors of the spurious tests could be like a TDD wall of shame, I like.
← PreviousPage 3 of 3