[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/biz/ - Business & Finance

View post   
File: 72 KB, 800x754, Output_screenshot_18.03.2018.png [View same] [iqdb] [saucenao] [google]
8429439 No.8429439 [Reply] [Original]

Hey /biz/, I'm creating a machine learning model to classify pictures and predict the market based on the ratio of bog/pepe/wojak/other posts.

Send me a picture to test it on. It's still a work in progress. If the pic you send is one of the above it'll get added to my training data.

Pic related, percent chance the model thinks it's in a category.

>> No.8429582
File: 578 KB, 3200x1680, jo.jpg [View same] [iqdb] [saucenao] [google]
8429582

>> No.8429657
File: 451 KB, 800x420, john_oliver_judged.png [View same] [iqdb] [saucenao] [google]
8429657

>>8429582
The bog/normal data definitely needs the most work.

>> No.8429679
File: 295 KB, 670x700, 1520795442578.png [View same] [iqdb] [saucenao] [google]
8429679

>> No.8429687
File: 15 KB, 259x194, guesswho.jpg [View same] [iqdb] [saucenao] [google]
8429687

>>8429439

>> No.8429715
File: 381 KB, 800x835, together.png [View same] [iqdb] [saucenao] [google]
8429715

>>8429679
Gonna add this to both the pepe and wojak datasets.

Deformed wojaks like that are harder I guess.

>> No.8429747
File: 282 KB, 800x599, merchant.png [View same] [iqdb] [saucenao] [google]
8429747

>>8429687
The outline of the merchant looks a lot like Pepe, actually.

>> No.8429804

Bump, give me more data fuckers.

>> No.8429811

>>8429439

>market sentiment analysis for images
>no sminem section

don't waste your time breh

>> No.8429839

>>8429811
Give me some smimen and i'll add him in.

>> No.8429853

>>8429439
congrats im gonna make a program that randomly post pink wojaks when the price is going up and green nujaks when the price is going down
hahaha faggot

>> No.8429883

>>8429853
Please do. I can grab the signal out of that.

>> No.8429897
File: 64 KB, 500x750, 1514925556528-biz.jpg [View same] [iqdb] [saucenao] [google]
8429897

>> No.8429923

>>8429883
sure you can kiddo
ill murky your signal more than you could ever imagine
i graduating top of my class of software engineering btw

>> No.8429940
File: 943 KB, 800x1200, got_it.png [View same] [iqdb] [saucenao] [google]
8429940

>>8429897

>> No.8429955

>>8429923
Got any good jobs lined up?
>Implying GPA matters.

>> No.8429971
File: 22 KB, 282x299, berry-valentine.jpg [View same] [iqdb] [saucenao] [google]
8429971

>> No.8429991
File: 85 KB, 1080x1349, 1521388623223.jpg [View same] [iqdb] [saucenao] [google]
8429991

>>8429439

>> No.8430015

>>8429923
>i graduating
sounds like its from a top university, anon

>> No.8430046
File: 270 KB, 800x848, fuck.png [View same] [iqdb] [saucenao] [google]
8430046

>>8429971

>> No.8430056
File: 48 KB, 396x382, 57B7B013-6B49-4CEA-A2FF-208089AA2C31.jpg [View same] [iqdb] [saucenao] [google]
8430056

>>8429439

>> No.8430060
File: 600 KB, 800x999, it_can_tell.png [View same] [iqdb] [saucenao] [google]
8430060

>>8429991

>> No.8430077
File: 352 KB, 800x771, thinks_its_a_wojak.png [View same] [iqdb] [saucenao] [google]
8430077

>>8430056

>> No.8430089
File: 73 KB, 1280x720, carlos.jpg [View same] [iqdb] [saucenao] [google]
8430089

>>8430060
you need special brrrrrrrrrrrrrrrrrrrrrap and trap categories for /biz/

>> No.8430115
File: 332 KB, 800x450, nothing.png [View same] [iqdb] [saucenao] [google]
8430115

>>8430089

>> No.8430141
File: 15 KB, 200x200, usdt.png [View same] [iqdb] [saucenao] [google]
8430141

>>8430115

>> No.8430155
File: 28 KB, 800x800, tether_confirmed_pepe.png [View same] [iqdb] [saucenao] [google]
8430155

>>8430141

>> No.8430174
File: 71 KB, 694x800, linky.jpg [View same] [iqdb] [saucenao] [google]
8430174

>>8430155
Now the real question since 50% of shitposts on /biz/ are about it.

>> No.8430192
File: 166 KB, 646x700, 1521103501572.png [View same] [iqdb] [saucenao] [google]
8430192

also i'm quite certain this wojak will be pepe

>> No.8430206
File: 1.12 MB, 1217x1685, Fibwojack.jpg [View same] [iqdb] [saucenao] [google]
8430206

>> No.8430221
File: 16 KB, 250x250, BD6023C4-A822-4157-95CC-C65D2C8F629E.jpg [View same] [iqdb] [saucenao] [google]
8430221

Big if true

>> No.8430230
File: 384 KB, 800x922, stiny_linky.png [View same] [iqdb] [saucenao] [google]
8430230

>>8430174
>Nothing

>> No.8430237
File: 107 KB, 761x915, 1474924931266.jpg [View same] [iqdb] [saucenao] [google]
8430237

>> No.8430248
File: 34 KB, 540x378, oold5ygcwlcy.jpg [View same] [iqdb] [saucenao] [google]
8430248

>> No.8430256
File: 27 KB, 472x472, 1520578596610.jpg [View same] [iqdb] [saucenao] [google]
8430256

>>8430230
>bog 0%
never ever above $1 confirmed

>> No.8430271
File: 51 KB, 1000x1000, 1417862633609.png [View same] [iqdb] [saucenao] [google]
8430271

>> No.8430279

>>8429439
can you make a program that counts how many times a coin is mentioned to see which coins are being talked about most and least on /biz/?

>> No.8430306

>>8430279
Already exists
https://twitter.com/cryptoballoffcl

>> No.8430311
File: 60 KB, 241x241, pard.png [View same] [iqdb] [saucenao] [google]
8430311

Can you make a trading bot that takes the latest 100 images on /biz/ and decides whether to buy or sell depending on the Pepe/Wojak ratio.

>> No.8430312
File: 217 KB, 800x866, greenie.png [View same] [iqdb] [saucenao] [google]
8430312

>>8430192
You're right, but I was close.

>> No.8430338
File: 563 KB, 800x560, def_confirmed.png [View same] [iqdb] [saucenao] [google]
8430338

>>8430248

>> No.8430349

would it be better for the machine to concentrate on the top 100 used images of pepe and wojak? Pareto principle.

>> No.8430352
File: 205 KB, 800x800, baby_pepe_wojak.png [View same] [iqdb] [saucenao] [google]
8430352

>>8430311
That's the idea.

>> No.8430368

>>8430349
Oops didn’t see other reply.

>> No.8430374

>>8430349
The images often get edited slightly, but I may weigh those images more based on how often they appear. I need a lot more training data.

>> No.8430404
File: 325 KB, 904x583, 1519487972459.jpg [View same] [iqdb] [saucenao] [google]
8430404

See if it recognizes this.

>> No.8430423
File: 9 KB, 225x224, 1517849062966.jpg [View same] [iqdb] [saucenao] [google]
8430423

Here's an easier one.

>> No.8430438
File: 462 KB, 800x515, kamikaze_pepe_judged.png [View same] [iqdb] [saucenao] [google]
8430438

>>8430404

>> No.8430441
File: 501 KB, 960x720, 1521069357156.jpg [View same] [iqdb] [saucenao] [google]
8430441

Here's a bit harder.

>> No.8430444

>>8430404
After I threw it in azure's computer vision API

{ "tags": [ "plane", "airplane", "flying", "skiing", "air", "small", "smoke", "red", "jet", "blue", "snow", "white", "man", "riding" ], "captions": [ { "text": "a plane flying in the air", "confidence": 0.5739788 } ] }

>> No.8430452
File: 40 KB, 485x443, 1521351708059.jpg [View same] [iqdb] [saucenao] [google]
8430452

>> No.8430458
File: 77 KB, 900x900, 1514739638927.jpg [View same] [iqdb] [saucenao] [google]
8430458

>>8430438
Impressive.

>> No.8430466
File: 194 KB, 800x796, wrong8.png [View same] [iqdb] [saucenao] [google]
8430466

>>8430423

>> No.8430481
File: 629 KB, 800x600, furry_wojak_judged.png [View same] [iqdb] [saucenao] [google]
8430481

>>8430441
Seems like the change in outline throws it off.

>> No.8430495
File: 295 KB, 800x730, judged_grug.png [View same] [iqdb] [saucenao] [google]
8430495

>>8430452

>> No.8430501
File: 65 KB, 466x615, 1518919577788.jpg [View same] [iqdb] [saucenao] [google]
8430501

>>8430481
Strange that it recognized the plane one so accurately.

Maybe it was the coloring.

>> No.8430512
File: 345 KB, 600x600, 56F89342-D533-4E63-BD0D-C22BF0C81EA2.png [View same] [iqdb] [saucenao] [google]
8430512

>> No.8430516
File: 479 KB, 800x800, 100_percent_pepe.png [View same] [iqdb] [saucenao] [google]
8430516

>>8430458

>> No.8430518
File: 27 KB, 478x380, C9ZlmpwWsAEP7ff.jpg [View same] [iqdb] [saucenao] [google]
8430518

Does it false-positive smug anime faces as wojaks?

>> No.8430537

>>8429439
The pictures are a response to the market movements, so unless you implicitly conjecture that /biz/ and the markets are an effective feedback loop, you're basing predictions off of reactions to past events which makes no sense.

>> No.8430541
File: 7 KB, 184x184, smug_face_anime_girl_steam_avatars.jpg [View same] [iqdb] [saucenao] [google]
8430541

>>8430516
Very nice.

>> No.8430542
File: 88 KB, 280x291, 1498261440064.png [View same] [iqdb] [saucenao] [google]
8430542

Try this.

>> No.8430543
File: 922 KB, 800x1055, tritium.png [View same] [iqdb] [saucenao] [google]
8430543

>>8430501
Yeah, guessing color. Not a hint of green there.

Post more bogs fuckers.

>> No.8430559
File: 399 KB, 800x635, almost_wojak.png [View same] [iqdb] [saucenao] [google]
8430559

>>8430518
Almost.

>> No.8430581
File: 245 KB, 800x831, FOR_SURE.png [View same] [iqdb] [saucenao] [google]
8430581

>>8430542
Definetly a wojak

>> No.8430582

>>8430537
This, the pictures usually imply some sort of emotion behind it.

I.e. there's a difference between a suicidal pink wojack and a green wojack with arrows flying behind him

>> No.8430589

>>8430582
Yeah. I'm probably gonna add more subcategories.

desu my plan is less to trade with it and more to see if the ratios actually correlate with market movements over time, just as a side project.

>> No.8430609

>>8430589
I see

>> No.8430615
File: 52 KB, 500x555, incoming-call-verizon-wireless-bogdanoff-reject-call-with-message-14385930.png [View same] [iqdb] [saucenao] [google]
8430615

Try this on for size

>> No.8430622
File: 1.01 MB, 2034x2940, benedict-cumberbatch.jpg [View same] [iqdb] [saucenao] [google]
8430622

Does he false-positive as a bog?

>> No.8430636

>>8430615
Green must have thrown it off.

Alright, give me a minute. Going to re-train it.

>> No.8430689
File: 235 KB, 800x888, strang.png [View same] [iqdb] [saucenao] [google]
8430689

>>8430615
Shit, forgot image.>>8430622
>>8430622
I'll do it next.

>> No.8430718
File: 1005 KB, 800x1156, confirmed_cumber.png [View same] [iqdb] [saucenao] [google]
8430718

>>8430622
Confirmed descended from ancient bloodlines

>> No.8430766
File: 27 KB, 400x400, 1515263652097.jpg [View same] [iqdb] [saucenao] [google]
8430766

>> No.8430790
File: 19 KB, 400x400, h6nqzAAq_400x400.jpg [View same] [iqdb] [saucenao] [google]
8430790

>> No.8430984

I think that's it for tonight, thanks for the data!

May return later this week with more refined models.

>> No.8431297
File: 119 KB, 500x480, wojak_wilde.png [View same] [iqdb] [saucenao] [google]
8431297

>>8430046
Well, it's pink.

How does your algorithm work, OP? Do you detect the most common color in the picture? Or maybe feed the bot some hashes of common images (which I would find very troublesome)?

>> No.8431435
File: 38 KB, 600x600, 20f947c0a41e6ed495867441e9d2c16c.jpg [View same] [iqdb] [saucenao] [google]
8431435

add mutts to algo

>> No.8431440

>>8431297
Neural network.

Just multiple layers.

The whole point is I don't know how it works. I just feed it the training data.

>> No.8431508
File: 138 KB, 800x600, nolonger.jpg [View same] [iqdb] [saucenao] [google]
8431508

>>8429439
how about dis?

>> No.8431566
File: 8 KB, 229x220, index.png [View same] [iqdb] [saucenao] [google]
8431566

>> No.8431592
File: 59 KB, 512x512, 1519275219019.jpg [View same] [iqdb] [saucenao] [google]
8431592

>> No.8431614
File: 56 KB, 409x395, 1521185698382.jpg [View same] [iqdb] [saucenao] [google]
8431614

>> No.8431616
File: 278 KB, 441x429, 312fj.png [View same] [iqdb] [saucenao] [google]
8431616

>>8429439

>> No.8431644
File: 1.26 MB, 960x1280, IMG_6662.png [View same] [iqdb] [saucenao] [google]
8431644

>>8429439

>> No.8431687

>>8429439
I already did this, it basically parses OP images in the catalog and returns a coefficient which evaluates overall market feeling

As a friendly pointer, bog posts provide no meaningful insight or data

>> No.8431786

>>8429439
You are what Clif High aspires to be.

>> No.8431819
File: 130 KB, 893x1024, 1521273698344.jpg [View same] [iqdb] [saucenao] [google]
8431819

>> No.8431830
File: 47 KB, 720x720, 1519535910600.jpg [View same] [iqdb] [saucenao] [google]
8431830

>> No.8431850

>>8431687
Nice; what features/images did you end up parsing out as important?

If you want to share, ofc.

>> No.8432363
File: 209 KB, 905x905, 5B420467-0E22-4105-AF38-FCB26E0AFDDD.jpg [View same] [iqdb] [saucenao] [google]
8432363

Rolling

>> No.8432553

How large is your current data set and how long did you spend preparing it?

>> No.8432625
File: 888 KB, 3200x3200, IMG_2141.jpg [View same] [iqdb] [saucenao] [google]
8432625

make the bot run every 30 minutes or so and have it post the results to a private telegram channel that us /biz/lets can see. this is an interesting project desu

>> No.8432855

>>8432553
~100 images each. I need more. Not very long.

>> No.8432886

>>8429439
Share the link of the GitHub repository OP. Once you are done

>> No.8434034
File: 1.09 MB, 1037x1037, 37d.png [View same] [iqdb] [saucenao] [google]
8434034

Try this