This page is not created by, affiliated with, or supported by Slack Technologies, Inc.
- # aleph (5)
- # beginners (92)
- # cider (37)
- # cljs-dev (38)
- # cljsjs (2)
- # cljsrn (3)
- # clojure (50)
- # clojure-berlin (1)
- # clojure-canada (3)
- # clojure-dusseldorf (4)
- # clojure-france (1)
- # clojure-germany (1)
- # clojure-italy (7)
- # clojure-nl (21)
- # clojure-spec (2)
- # clojure-uk (106)
- # clojurescript (165)
- # code-reviews (1)
- # community-development (3)
- # cursive (5)
- # datomic (13)
- # editors (12)
- # emacs (3)
- # figwheel-main (141)
- # fulcro (28)
- # graphql (1)
- # immutant (1)
- # jobs (1)
- # jobs-discuss (5)
- # midje (8)
- # nrepl (3)
- # off-topic (28)
- # onyx (4)
- # re-frame (21)
- # reagent (70)
- # ring (2)
- # ring-swagger (9)
- # shadow-cljs (18)
- # spacemacs (6)
- # specter (23)
- # tools-deps (21)
@jasonbell Do you do ML stuff in Clojure, or is that an area where you use Python or similar?
I need to brush up on some Python stuff. Even to get some of that over marketed Tensorflow stuff.
@jasonbell you are an old git...but one, that's got nothing to do with using Java, two Python is older than Java as a language and three you're not as old a git as me! 😂
I had a look at a ML "Hello World" style app in Clojure (categorising tulips or something?). It broke my mind. I could see what each individual function did easily enough, I just couldn't see why putting them all together produced the expected result
Ah, the joys of receiving letters obviously intended for identity theft. Who to send a lovely GDPR letter to?
I need to do more reading on ML, but I suspect it involving a lot more mathsy stuff than I am used to counts against me
import supermegaacemllibrary input_file = csv("./some-input") create_business_value(input_file)
even in spark/scala with some kaggle datasets it was a bit like "okay, so I've trained what to do what? I think I'm datascience-ing but everything is a black box"
@alex.lynham I found it alarming how many books/tutorials and so on only ever focused on the training and never once showed you how to do a prediction.
If it makes you feel better, even if you understand the maths neural networks/deep learning are still black boxes
Hence everyone at NI conferences gets upset when I say don't use them after hearing every other speaker say "use them"
@otfrom I liked that article you posted about deep learning (or layers of it at least) basically coming down to polynomial regression (though I think it was just feed-forward NN and not recursive ones)
I reckon if you get really good you probably have a vague idea of what all the tuning params do to the shape of the function underneath
but it comes down to the fact that it's complexity on top of complexity. Do most of us really understand what a Clojure function translates to in terms of assembly instructions? Don't reckon so
@peterwestmacott I wish I could. I know a few colleagues who are down there and my wife would be if she wasn't off to a hen do in a short while
I was going to, and then I was like... I'm not spending money to protest when trump can f--k off for free
this is cool to get an intuitive feel for things https://playground.tensorflow.org/#activation=tanh&batchSize=10&dataset=circle&regDataset=reg-plane&learningRate=0.03&regularizationRate=0&noise=0&networkShape=4,2&seed=0.82645&showTestData=false&discretize=false&percTrainData=50&x=true&y=true&xTimesY=false&xSquared=false&ySquared=false&cosX=false&sinX=false&cosY=false&sinY=false&collectStats=false&problem=classification&initZero=false&hideText=false
"In statistics, polynomial regression is a form of regression analysis in which the relationship between the independent variable x and the dependent variable y is modelled as an nth degree polynomial in x." ... Where is Wikipedia's ELI5? What makes a variable independent or dependant? WTF is an "nth degree polynomial"? What makes it "in x"? ML stuff I've read seems to assume I already know this sort of thing instead of starting from basics there too
well, you're trying to fit a function to existing data points. Linear regression fits a function y = ax + b, by estimating a and b
polynomial is a generalization, you can have higher order (math) functions: say y = ax^2 + bx + c
that's only with one variable x, you can have more than one x for more than one feature. The more features and the higher the order, the more parameters you have to optimize
Seems like a scary path to take haha. I don’t think i’ve ever really worked for an ethical company in london.
mmm... I just moved a transducer and a reduce from a core.async channel and slapped the same code into the guts of a
. I do love that transducers make that easy
Has anyone got a good example that they can point me at of looping over a lazy-seq, in order to use each item until a condition is met?
I have a lazy-seq of DateTime objects and I want to do something for each one until I get to a particular date...
here's my lazy-seq getting made, where 'jt' is clojure.java-time:
(def timeline (jt/iterate jt/plus (jt/local-date 2004 1 1) (jt/days 1)))
I mean it looks as though take-while will return a collection of all the items that comply to the predicate... I'd still need to process that collection
I am going out of my mind... Lazy Sequence of clojure.java-time LocalDates that I need to iterate over until I reach a specified date, but I can't compare the items that come off the lazy sequence 'cos they aren't realised so they can't be compared with the LocalDate I am creating for the specified end date.
Turns out that a seq of LocalDate objs from 2004-01-01 to now with a step of 1day doesn't blow the stack so I can__ take-while and then doseq