* Maʻemaʻe no ke aʻo ʻana i ka Machine Learning, ʻoiaʻiʻo. Ma lalo o ka nānā iki ʻana o kāna wahine aloha.
ʻAʻohe paha noi e like me ka maʻalahi i ka pae o ka spinal reflexes e like me Tinder. I mea e hoʻohana ai, pono ʻoe i hoʻokahi manamana lima e swipe a me kekahi mau neurons e koho i nā kaikamahine a i ʻole nā kāne āu e makemake ai. ʻO kahi hoʻokō kūpono o ka brute force ma ke koho pālua.
Ua hoʻoholo wau he ala maikaʻi kēia e loaʻa ai kahi manaʻo iki no ke aʻo ʻana i ka mīkini ma kahi kāleka kiʻi hou. ʻO nā mea a pau i koe, ʻo ka wehewehe ʻana i kaʻu wahine ʻaʻole pono iaʻu i kahi wahine momona hou, a ke aʻo nei au i nā neural network.
He aha ka pilikia me nā pūnaewele pili?
Aia kekahi kumu waiwai - Ashley Madison. Specific, me ka slogan "He pōkole ke ola. E pili." ʻO ka poʻe nui ka poʻe kāne male e ʻimi ana i kahi hihia ma ka ʻaoʻao. He leʻaleʻa hoʻi ka monetization - ma kahi o ka "hoʻolilo i nā helu e makemake a kākau," noi lākou i $ 19 e holoi i ka moʻokāki o ka mea hoʻohana me ka ʻole o ka ʻike.
I ka makahiki 2015, ua lele maoli ka pūnaewele a 60 GB o ka ʻikepili pilikino i hoʻokuʻu ʻia i loko o ka waihona lehulehu. Ma waho aʻe o nā ʻohana i luku ʻia, ua hāʻawi kēia leak i nā ʻike hoihoi i ka poʻe loiloi. Manaʻo mau wau he nui nā kāne ma nā pūnaewele pili, akā i kēia hihia ua lilo ia i mea hoihoi loa. ʻO Annalee Newitz ka mea kākau moʻolelo Ua ʻike ʻia ʻo 5 miliona mau mea hoʻohana, ʻo 12 wale nō i like me nā moʻolelo kaikamahine maoli a hoʻohana mau ʻia. ʻO ke koena he mau bots wale nō i kamaʻilio me nā kāne malihini.
ʻAʻole maʻamau ka manaʻo nui i nā moʻolelo kāne no kēia kumuwaiwai, akā no ka hapa nui o nā pūnaewele pili. ʻIke wau he nui ka poʻe i hālāwai me kēia kūlana kūpono ʻole, ke hoʻolālā pono ʻoe i kahi kamaʻāina, akā pono ke kaikamahine e hoʻopaʻa inoa. E haʻalele kākou i ka maikaʻi o kēia lehulehu o ka poʻe pā, akā ʻo ka ʻoiaʻiʻo ʻaʻole hiki ke hōʻole ʻia ke neʻe ʻana o ke koena o ka hoʻolako a me ke koi i nā kaikamahine.
Hiʻona Tinder
ʻO ka mea hoʻoikaika kino kūpono i ka pilina kāne
ʻO ka hiʻohiʻona nui o kēia paepae ke kumu kūʻai haʻahaʻa no kēlā me kēia hoa. Ua lawa ka coincidence o ʻelua swipes a ke kamaʻilio nei ʻoe me kahi kanaka hoihoi. ʻO ka pilikia, ʻo ka like ʻole o ke kāne like ʻole e alakaʻi i ka ʻoiaʻiʻo e loaʻa i ka hapa nui o nā kaikamahine he mau pāʻani i kēlā me kēia lā. ʻO ia hoʻi, ʻaʻohe o lākou manawa e hoʻolohe ai iā ʻoe ma waena o nā moho ʻē aʻe.
He mea maopopo loa ka manaʻo o ka paepae i ka manawa liʻiliʻi e loiloi i ka honua hohonu o ke kanaka mai ka hoʻokahi a me ka hapa ʻelua o ka nānā ʻana i kahi kiʻi i loko o kahi ʻauʻau kai a i ʻole ke kaʻa ʻana i kahi kaʻa ʻoniʻoni. No laila, inā ʻaʻole ʻoe e nānā akua i kāu mau kiʻi, ʻaʻohe āu koho akā e hoʻonui i kou manawa ma ka lawe ʻana. ma kekahi mau ano. Ma ka waiho wale ʻana, e hoʻoikaika mākou i ka ikaika a lawe i ka leo e hoʻonui i ko mākou manawa o ka holomua hānau. No ka mea, i kekahi manawa pono ʻoe e hoʻopilikia ʻia e ka meaʻai a me ka hiamoe, a ua kaupalena ʻia nā swipes, makemake paha ʻoe e koho ʻo automation i nā kaikamahine a i ʻole nā kāne i kūpono i kou makemake. ʻO nā ʻulaʻula pōkole a i ʻole nā ʻulaʻula lōʻihi - aia iā ʻoe.
ʻOhi ʻikepili
ʻO ka mea mua, pono ʻoe i ka nui o ka ʻikepili no ka pololei maʻamau. Ua ʻike ka poʻe i ʻike i ke aʻo ʻana i ka mīkini i ka paʻakikī o ka hana ʻana i kahi ʻikepili i hōʻiliʻili ʻia a hoʻopaʻa inoa ʻia. ʻO ka manaʻo, e kūpono nā kumuwaiwai like ʻole ma ke ʻano he kumu ʻikepili, ʻo ia ʻo Instagram a i ʻole nā mea pūnaewele ʻē aʻe. Akā ʻoi aku ka maikaʻi o ka hoʻomaʻamaʻa ʻana i kēlā mau laʻana e hana ai ka pūnaewele i ka wā e hiki mai ana.
E lawe kākou i ka waihona i kumu . Loaʻa i ka lehulehu nā kiʻi a Tinder, akā ua kaupalena ʻia ka hana "like". No laila, pono e unuhi i nā mea ola a pau i loko o ka radius a makaʻala pono iā lākou. Pono mua ʻoe e hoʻohana i kahi palapala maʻalahi:
from skimage.io import imread, imsave, imshow, show
import matplotlib.pyplot as plt
import pynder
from helpers import get_access_token, get_login_credentials
from io_helper import save_image
email, password, FBID = get_login_credentials()
FBTOKEN = get_access_token(email, password)
session = pynder.Session(facebook_token=FBTOKEN)
while True:
users = session.nearby_users()
for user in users:
photos = user.get_photos()
print("Fetched user photos..")
for photo in photos:
print(photo)
image = imread(photo)
imshow(image)
show()
input_string = "Write 1 to like. Write 2 to dislike."
ans = str(input(input_string)).lower()
if ans == "1":
save_image(image, photo, True)
else:
save_image(image, photo, False)
E ʻae iā ʻoe e hōʻailona i ka ʻikepili i ka wikiwiki me nā pihi ʻelua wale nō. Aia ka lua kī ma ka ʻoiaʻiʻo ua haki ʻia ka waihona werkzeug i hope a pono e koi ʻia e hoʻohaʻahaʻa. A i ʻole, hoʻolei ia i kēia hewa.
Traceback (most recent call last):
File "img_scrape.py", line 4, in <module>
from helpers import get_access_token, get_login_credentials
File "/home/someone/tmp/TinderAutomation/helpers.py", line 1, in <module>
import robobrowser
File "/home/someone/tmp/TinderAutomation/venv/lib/python3.6/site-packages/robobrowser/__init__.py", line 3, in <module>
from .browser import RoboBrowser
File "/home/someone/tmp/TinderAutomation/venv/lib/python3.6/site-packages/robobrowser/browser.py", line 8, in <module>
from werkzeug import cached_property
ImportError: cannot import name 'cached_property'
No laila, ma requirements.txt pono ʻoe e kākau iā Werkzeug ==0.16.1. A laila e haʻalele.
ʻO ka pilikia ʻelua ka loaʻa ʻana o kēia hōʻailona. ʻAʻole i hana ke ʻano maʻamau mai ka waihona iaʻu, akā hiki iaʻu ke kiʻi mai ka console developer. No ka hana ʻana i kēia, hele i a wehe i ka pane i ka POST noi ma . I loko mākou e ʻimi ai no 'access_token'. No kekahi kumu ʻaʻole ia i hana i ka manawa mua, akā ua loaʻa iaʻu a paʻakikī i loko o ka palapala.
Nā koi ʻikepili
Aia kekahi mau koi nui no ka ʻikepili aʻo mīkini:
- Ka lawa pono
- Kaulike
- Ka nui
ʻO ka lawa i kēia hihia e pono ai ma kahi o 10000 mau kiʻi e kūkulu i kahi kumu hoʻohālike kūpono. ʻAe, nui kēlā. ʻO kēia ke kumu e like ai nā lawelawe , kahi no ka uku hiki iā ʻoe ke hāʻawi i ka markup o kāu dataset i nā poʻe ʻē aʻe. Ma ka ʻaoʻao ʻē aʻe, makemake maoli ʻoe i kāu bot e makemake i nā kaikamahine ʻAsia maikaʻi i ka maka mahina a i ʻole nā kaikamahine nani like me nā aʻa India? Eia nō naʻe, pono e hōʻike ke kumu hoʻohālike i kou ʻono.
ʻAʻohe pilikia me nā ʻano like ʻole; hōʻike ʻia nā kiʻi āpau mai nā kihi like ʻole a me nā kukui. I nā aniani, nā lole, nā lole ʻauʻau a me nā lole ski. Hiki mai kekahi pilikia me ka like o ka waihona. ʻO ke kūpono, ke kau inoa mākou i kā mākou laʻana, pono e loaʻa i nā ʻāpana like. Inā hoʻopau ʻoe i kahi ʻikepili "skewed", pono ʻoe e hoʻoheheʻe me nā kiʻi mai nā kumu ʻē aʻe. Pono ʻoe e hoʻohui i nā mea hoihoi, a i ʻole, e hoʻoholo ʻoe iā lākou ma muli o ka hopena markup. Ua loaʻa iaʻu kahi mea ma kahi o 60% nani. ʻAʻole wau koho nui, a i ʻole he laki wale nō a nui nā kaikamahine uʻi a puni.
ʻAʻole hoʻi au e hōʻemi i ke kuhiakau he nui nā bots i waena o lākou. Hoʻomaʻamaʻa mākou i kahi bot e makemake i nā bots ʻē aʻe. Aia kekahi mea hoʻohenehene i kēia.
Ka hoʻoponopono ʻikepili
Loaʻa iā mākou nā kiʻi i hoʻopaʻa ʻia, akā hui pū lākou. ʻO ke ao, ka pō, mai hope a me nā mea ʻē aʻe. Me ka minamina, maopopo iaʻu ʻaʻole maikaʻi loa ke aʻo ʻana mai nā kiʻi mai ka ʻaoʻao ʻē aʻe, no ka mea, ʻaʻole kūlike ka laʻana. No laila, ʻo ka koho maikaʻi loa ʻo ia ka hoʻohana ʻana i nā maka ma ke ʻano he hōʻailona kuhikuhi o ka "cuteness." Eia nō naʻe, no mākou, e like me nā primates ʻē aʻe, he ʻano kī kēia.
No laila, e hoʻohana kākou . He algorithm maikaʻi kēia e hiki ai iā ʻoe ke ʻimi i nā maka i nā kiʻi me ka haʻahaʻa haʻahaʻa o nā hewa maikaʻi ʻole.
Ua wehewehe nui ʻia kēia ma ka manual
I ka pae aʻe, ma hope o nā helehelena wale nō i loko o ka laʻana, pono e wehe i ke kala. ʻO ka ʻoiaʻiʻo, ʻaʻole pono ʻoe e koho ma waena o ka denizen uliuli nani o Pandora a i ʻole ka nani ʻili ʻōmaʻomaʻo.
I ka poʻe Hue, ʻaʻole nui ka ʻili o ka ʻili i nā loiloi o ka nani.
No laila, pono e hoʻomaʻamaʻa i ka hana o ka neural network a waiho wale i ka grayscale.
Hale hoʻohālike
Makemake wau e ʻōlelo koke me ka loaʻa ʻole o kahi kāleka wikiō maikaʻi a me CUDA, ʻaʻole paha ʻoe e loaʻa i kahi kumu hoʻonaʻauao i ka manawa kūpono. No laila, e ʻimi koke i ka helu ʻana i nā ao kūikawā a i ʻole ka hoʻohana ʻana i ka python-CUDA.
Ua lawe au i kahi hiʻohiʻona kumu ʻekolu-layer mai ka mea kākau o ka waihona a, kahaha, ua hōʻike ʻo ia i ka pololei ma kahi o 72%, he hopena maikaʻi loa ia.
model = Sequential()
model.add(Convolution2D(32, 3, 3, activation='relu', input_shape=(img_size, img_size, 3)))
model.add(MaxPooling2D(pool_size=(2,2)))
model.add(Convolution2D(32, 3, 3, activation='relu'))
model.add(MaxPooling2D(pool_size=(2,2)))
model.add(Convolution2D(64, 3, 3, activation='relu'))
model.add(MaxPooling2D(pool_size=(2,2)))
model.add(Flatten())
model.add(Dense(128, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(2, activation='softmax'))
adam = optimizers.SGD(lr=1e-4, decay=1e-6, momentum=0.9, nesterov=True)
model.compile(loss='categorical_crossentropy',
optimizer= adam,
metrics=['accuracy'])
Inā loaʻa kahi laʻana maikaʻi, a laila lawa paha ka loaʻa ʻana o kahi kumu hoʻohālike kūpono.
E hoʻomaka kākou i ka bot
Mahalo i ka mea kākau o ka waihona no kahi koho mākaukau no ka hoʻāʻo wikiwiki ʻana i ka manaʻo. ʻOiaʻiʻo, hana maikaʻi loa ia i ka mana kumu a hiki ke hoʻokuʻu ʻia ma kā mākou . ʻAʻole hiki ke hoʻomaʻamaʻa; i kēia manawa ʻaʻole mākou e hāʻawi i nā mīkini virtual me ke kākoʻo CUDA no ka helu ʻana, akā hiki iā ʻoe ke holo i kahi mea 24/7 me ka pilikia ʻole. He māmā loa ka bot, no laila ʻoi aku ka maikaʻi o ka lawe ʻana i ka uku e uku ai i nā kumuwaiwai i hoʻohana ʻia.
Nā hualoaʻa
Manaʻo wau he nani loa wau. A he honua waiwai koʻu i loko. Ua loaʻa iaʻu kahi mea e like me 13 mau pāʻani i loko o hoʻokahi hola. Eia kekahi, ua kākau mua nā kaikamahine i kekahi mau manawa.
ʻO ka hopena, ua hoʻopau mākou me nā kamaʻilio maikaʻi loa, kahi aʻu i ʻōlelo ai ua hele mai wau e pāʻani wale me ka aʻo ʻana i ka mīkini a me ka hōʻailona ʻikepili. Ua hoihoi loa kekahi o na kaikamahine, no ka mea, he haku hana oia. Aia ka manaʻo ikaika e heluhelu ʻo ia i kēia pou ma Habré. Manaʻolana wau e mālama ʻo Oksana i koʻu inoa ʻole. 🙂
*ʻale i ka wāwae a ʻōlelo ʻo ia aloha
He wahi liʻiliʻi e pili ana i ka ʻaoʻao pono o ka hihia
ʻO kaʻoiaʻiʻo, ʻaʻole wau makemake i ka manaʻo holoʻokoʻa o ka robotizing pilina ma waena o nā kāne a me nā kaikamahine. Aia kekahi mea kūpono loa i ke kiola ʻana i kou ʻaʻahu ma luna o nā poʻohiwi o kahi malihini anu e kū hoʻokahi ana. A i ʻole e hoʻokokoke i kahi kaikamahine uʻi i loko o kahi cafe kauwela a inu kofe pū. E hele i waho ma hope o nā mea nānā.
Ua puni ke kauwela. ʻO ka manawa kēia e kamaʻāina ai.
Source: www.habr.com