*Dalla-dalla don koyan Injin Learning, ba shakka. A k'ark'ashin kallon rashin gamsuwa da matarsa abar so.
Wataƙila babu wani aikace-aikace mai sauƙi zuwa matakin jujjuyawar kashin baya kamar Tinder. Don amfani da shi, kuna buƙatar yatsa ɗaya kawai don shafa da ƴan neuron don zaɓar 'yan mata ko mazan da kuke so. Ingantacciyar aiwatar da ƙwaƙƙwaran ƙarfi a zaɓi biyu.
Na yanke shawarar cewa wannan zai zama hanya mai kyau don samun ɗan jin daɗin koyon injin akan sabon katin zane. Abin da ya rage shi ne in bayyana wa matata cewa bana bukatar sabuwar mace mai kiba, kuma ina horar da hanyoyin sadarwa ne kawai.
Mene ne matsalar sadarwar sadarwar zamantakewa?
Akwai irin wannan albarkatun - Ashley Madison. Musamman, tare da taken “Rayuwa gajeru ce. Ku yi mu'amala." Babban masu sauraro mazan aure ne masu neman al'amari a gefe. Har ila yau, samun kuɗin shiga yana da daɗi - ban da daidaitattun "kashe maki don so da rubutawa," sun nemi $19 don share asusun mai amfani ba tare da wata alama ba.
A cikin 2015, rukunin yanar gizon ya bazu kuma 60 GB na bayanan sirri ya leka cikin jama'a. Baya ga iyalai da yawa da aka lalata, wannan ledar ta ba da bayanai masu ban sha'awa ga manazarta. A koyaushe ina zargin cewa akwai ƙarin maza da yawa akan rukunin yanar gizo, amma a wannan yanayin ya zama mai ban sha'awa sosai. 'Yar jarida Annalee Newitz ya gano cewa daga cikin masu amfani da miliyan 5, kawai 12 sun kasance daidai da asusun 'yan mata na ainihi kuma ana amfani da su akai-akai. Sauran bots ne kawai waɗanda ke tattaunawa da baƙi maza.
Irin wannan ra'ayi game da asusun maza ya zama na hali ba kawai don wannan albarkatu ba, har ma ga yawancin sauran rukunin yanar gizon. Na tabbata cewa mutane da yawa sun fuskanci wannan yanayin rashin adalci, lokacin da za ku yi shiri a hankali, amma yarinyar kawai tana buƙatar yin rajista. Bari mu bar ingancin wannan taron jama'a na magoya baya, amma gaskiyar ba za a iya musantawa ba cewa ma'auni na wadata da buƙatu yana nunawa a fili ga 'yan mata.
Siffar Tinder
Maƙasudin ƙaƙƙarfan ƙarfi a cikin alaƙar jinsi
Babban fasalin wannan dandali shine ƙarancin farashi ga kowane wanda aka sani. Daidaiton ɓangarorin biyu ya isa kuma kun riga kun yi sadarwa tare da mutum mai yuwuwar ban sha'awa. Matsalar ita ce rashin daidaiton jinsi ɗaya yana haifar da gaskiyar cewa yawancin 'yan mata za su sami da yawa ashana kowace rana. Wannan yana nufin cewa da alama ba za su sami lokaci don kula da ku a cikin sauran 'yan takara ba.
A bayyane yake cewa dandamali yana ba da damar kaɗan don tantance zurfin duniyar ciki na mutum daga kallo ɗaya da rabi na biyu a hoto a cikin rigar iyo ko tuƙin mota mai launi na gaye. Don haka, idan ba ku yi kama da allahntaka kawai a cikin hotunanku ba, ba ku da wani zaɓi illa ƙara damarku ta hanyar ɗauka. a wasu nau'ikan. A taƙaice, za mu zage damtse kuma za mu ɗauki ƙara don ƙara yuwuwar samun nasarar kiwo. Tun da a wasu lokuta kuna buƙatar shagaltar da ku ta abinci da barci, kuma ƙwanƙwasa ba ta da iyaka, tabbas za ku fi son cewa sarrafa kansa ya zaɓi 'yan mata ko maza waɗanda suka dace da abubuwan da kuke so. Jajayen jajayen gajere ko dogon brunettes - ya rage na ku.
Tattara bayanai
Da farko, kuna buƙatar bayanai da yawa don daidaito na al'ada. Duk wanda ya ci karo da koyon na'ura ya san yadda zai yi wahala a samar da saitin bayanai da aka tattara daidai da lakabi. A ka'ida, duk wani abu makamancin haka zai dace da matsayin tushen bayanai, ya zama Instagram ko sauran hanyoyin sadarwar zamantakewa. Amma yana da kyau a horar da waɗannan samfurori waɗanda hanyar sadarwar za ta yi aiki a nan gaba.
Bari mu dauki ma'ajiyar a matsayin tushe . Hotunan Tinder koyaushe suna samuwa a bainar jama'a, amma aikin "kamar" ya riga ya iyakance. Sabili da haka, wajibi ne a cire dukkan abubuwa masu rai a cikin radius kuma a hankali su yi alama. Da farko kuna buƙatar amfani da ingantaccen rubutun rubutu:
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)
Zai ba ka damar yin alama da saitin bayanai da sauri tare da maɓallai biyu kawai. Maɓallin maɓalli ya ta'allaka ne a cikin gaskiyar cewa ɗakin karatu na werkzeug ya karya daidaituwar baya kuma dole ne a tilasta masa rage darajarsa. In ba haka ba yana jefa wannan kuskure.
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'
Don haka, a cikin buƙatun.txt kuna buƙatar rubuta Werkzeug==0.16.1. Sannan zai tashi.
Matsala ta biyu ita ce samun wannan alama. Hanyar daidaitaccen tsari daga ma'ajiyar ba ta yi aiki a gare ni ba, amma na sami damar samun ta daga na'ura mai haɓakawa. Don yin wannan, je zuwa kuma fitar da amsa ga buƙatar POST a ciki . A ciki muna neman 'access_token'. Don wasu dalilai bai yi aiki a karon farko ba, amma sai na same shi kuma na sanya shi cikin rubutun.
Bukatun saitin bayanai
Akwai mahimman buƙatu da yawa don saitin bayanan koyan inji:
- Wadata
- Daidaituwa
- Bambancin
Isasshen wannan yanayin yana buƙatar aƙalla hotuna 10000 don gina ingantaccen samfuri. Ee, hakan yayi yawa. Wannan shine ainihin dalilin da yasa ayyuka suke so , inda a kan kuɗi za ku iya ba da alamar alamar bayanan ku ga wasu mutane. A gefe guda, shin kuna son bot ɗin ku ya so kyawawan 'yan matan Asiya masu fuskantar wata ko daidai da kyawawan 'yan mata masu tushen Indiya? Duk da haka, samfurin ya kamata ya nuna dandano.
Babu takamaiman matsaloli tare da iri-iri; duk hotuna ana gabatar dasu daga kusurwoyi daban-daban da haske. A cikin tabarau, riguna, swimsuits da ski suits. Matsala na iya tasowa tare da daidaiton tsarin bayanan. Da kyau, lokacin da muka yiwa samfurin mu lakabi, yakamata ya ƙunshi kusan sassa daidai. Idan kun ƙare da saitin bayanan "skewed", dole ne ku shafe shi da hotuna daga wasu tushe. Kuna buƙatar ƙara ƙarin masu ban sha'awa, ko akasin haka, zaku ƙididdige su bisa sakamakon alamar. Na sami wani abu kusan 60% kyakkyawa. Ko dai ba ni da zaɓe, ko kuma na yi sa'a kuma akwai kyawawan 'yan mata da yawa a kusa.
Ba na kuma rage hasashe cewa akwai bots da yawa a cikinsu. Muna horar da bot wanda zai so sauran bots. Akwai abin ban mamaki a cikin wannan.
Sarrafa bayanai
Muna da tarin hotuna da aka yiwa alama, amma sun cakude sosai. Rana, da dare, daga baya da sauransu. Tare da nadama, na fahimci cewa koyarwa daga hotuna daga kusurwar baya ba zai yi tasiri ba musamman, tun da samfurin zai kasance mara daidaituwa. Sabili da haka, mafi kyawun zaɓi shine amfani da fuskoki azaman alamar nunin "kyakkyawa." Har yanzu, a gare mu, amma ga sauran primates, wannan mahimmin siga ce.
Saboda haka, bari mu yi amfani . Wannan kyakkyawan algorithm ne wanda ke ba ku damar nemo fuskoki a cikin hotuna tare da ƙaramin adadin kurakurai masu inganci na ƙarya.
An bayyana wannan dalla-dalla a cikin littafin
A mataki na gaba, bayan kawai fuskoki suna cikin samfurin, yana da ma'ana don cire launi. A gaskiya ma, da kyar za ku zaɓi tsakanin kyawawan shuɗi na Pandora ko kyan fata mai launin kore.
A cikin mutanen Hue, launin fata ba ya ba da gudummawa ga ƙima mai kyau.
Sabili da haka, yana da daraja sauƙaƙe aikin cibiyar sadarwar jijiyoyi kuma barin kawai launin toka.
Samfurin gini
Ina so in faɗi nan da nan cewa ba tare da katin bidiyo mai kyau da CUDA ba, da alama ba za ku sami samfurin horarwa a cikin isasshen lokaci ba. Don haka, nan da nan nufin ƙididdigewa a cikin gajimare na musamman ko amfani da python-CUDA.
Na ɗauki misali na asali mai nau'i uku daga marubucin ma'ajin kuma, abin mamaki, ya nuna daidaito na kusan 72%, wanda shine kyakkyawan sakamako.
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'])
Idan akwai samfuri mai kyau, to yana iya zama isa don samun samfurin aiki.
Bari mu kaddamar da bot
Godiya ga marubucin ma'ajin don zaɓin da aka shirya don gwada ra'ayin da sauri. A zahiri, yana aiki sosai a cikin sigar asali kuma ana iya ƙaddamar da ƙa'idar akan mu . Ba zai yiwu a horar da shi ba tukuna; a halin yanzu ba mu samar da injunan kama-da-wane tare da tallafin CUDA don ƙididdigewa ba, amma kuna iya gudanar da wani abu 24/7 ba tare da wata matsala ba. Bot ɗin yana da nauyi sosai, don haka zai fi riba a ɗauki jadawalin kuɗin fito da ke biyan albarkatun da aka yi amfani da su.
Результаты
Ina tsammanin ina da kyau sosai. Kuma ina da duniyar ciki mai arziki. Na sami wani abu kamar matches 13 a cikin sa'a guda. Bugu da ƙari, sau da yawa 'yan matan sun rubuta farko.
A sakamakon haka, mun ƙare tare da tattaunawa masu kyau, inda na ce na shigo ne kawai don yin wasa da koyon inji da lakabin bayanai. Daya daga cikin 'yan matan ta kasance mai matukar sha'awa, tun da ita kanta mai haɓakawa ce. Akwai ji mai ƙarfi cewa a ƙarshe za ta karanta wannan post akan Habré. Ina fatan Oksana za ta ci gaba da ɓoye sunana. 🙂
*tafad'a tana fad'in sannu
Kadan game da bangaren da'a na batun
A gaskiya, ba na son dukan ra'ayin na robotizing dangantaka tsakanin maza da 'yan mata. Akwai wani abu mai kyau game da jefa jaket ɗinku a kan kafaɗun baƙo mai sanyi wanda ke tsaye shi kaɗai. Ko kusanci wata kyakkyawar yarinya a cikin cafe rani kuma ku sha kofi tare. Fita daga bayan masu saka idanu riga.
Lokacin bazara yana ko'ina. Lokaci ya yi da za a saba.
source: www.habr.com