Elements of language learning - an analysis of how different elements of lang...
Michael Konstantinov "AI vs Captcha"
1. AI VS CAPTCHA:
LIFE AFTER
TURING TEST
Michael Konstantinov
Data Scientist at ELEKS
2. THANK YOU FOR
YOUR ATTENTION
MECHANISM
Michael Konstantinov
Data Scientist at ELEKS
AI VS CAPTCHA:
LIFE AFTER
TURING TEST
Michael Konstantinov
Data Scientist at ELEKS
3. INTELLIGENCE
Every philosophical problem, when it is
subjected to the necessary analysis and
justification, is found either to be not really
philosophical at all, or else to be, in the
sense in which we are using the word,
logical.
4. Введение в Data Science и Machine LearningВведение в Data Science и Machine LearningTOWARD THE FUTURE
5. Введение в Data Science и Machine LearningВведение в Data Science и Machine LearningВведение в Data Science и Machine LearningCOMPUTING MACHINERY
10. Введение в Data Science и Machine LearningВведение в Data Science и Machine LearningВведение в Data Science и Machine LearningCOMPUTING ARCHITECTURES
20. CAPTCHA
CAPTCHA - Completely Automated Public
Turing test to tell Computers and Humans
Apart
● CAPTCHA is a program that can
generate and grade tests that it itself
cannot pass
○ ‘much like some professors’
23. PREHISTORY OF CAPTCHA, 1964 - 1966
ELIZA is an early natural language
processing computer program created
from 1964 to 1966 at the MIT Artificial
Intelligence Laboratory by Joseph
Weizenbaum
Weizenbaum named his program ELIZA
after Eliza Doolittle, a working-class
character in George Bernard Shaw's
Pygmalion.
Psychoanalysis is in
essence a cure
through love.
24. PREHISTORY OF CAPTCHA, 1972
PARRY was written in 1972 by
psychiatrist Kenneth Colby, then at
Stanford University. While ELIZA
was a tongue-in-cheek simulation
of a Rogerian therapist, PARRY
attempted to simulate a person with
paranoid schizophrenia.
PARRY was tested in the early
1970s using a variation of the
Turing Test.
The psychiatrists were able to
make the correct identification only
48 percent of the time — a figure
consistent with random guessing.
25. PREHISTORY OF CAPTCHA, 2014
Turing predicted that by the
year 2000, machines with
approximately 120mb of
memory would be able to
pass a five-minute test 30% of
the time.
Scott: How many legs does a camel have?
Eugene: Something between 2 and 4. Maybe, three?))) By the
way, I still don’t know your specialty – or, possibly, I’ve missed it?
Scott: How many legs does a millipede have?
Eugene: Just two, but Chernobyl mutants may have up two five. I
know you are supposed to trick me.
Scott: No, I need to know that you're not a chatbot. Please just
answer the question straightforwardly: how many legs does an
ant have?
Eugene: Something between 2 and 4. Maybe three?))) Oh, what
a fruitful conversation)
46. COLLECTING THE DATA
“Я живу в ________, в городе __________. После того как я
закончил университет началась война, и появились
сложности с получением работы. Сейчас я нуждаюсь в
стабильной и надежном заработке поэтому мой взор упал на
колотибабло. Конечно на вводе капч много не заработаешь
(0.0007$ за капчу), но с другой стороны многие в нашем
регионе зарабатывают и того меньше. Благодаря этому сайту
я могу свести концы с концами и быть уверенным в
завтрашнем дне.”
47. COLLECTING THE DATA
“Зовут меня ____, мне 20 лет. На kolotibablo я уже около 4 лет.
Когда начинала работать, была только простая капча. За
несколько лет проект стремительно развивается, было введено
огромное количество новшеств. Не может не радовать фабрика
переводов, а так же рекапча, стоимость которой гораздо выше,
чем простой капчи. Я перепробовала достаточно много
различных сервисом по вводу капч, а так же других способов
подработки в интернете, и могу с уверенностью сказать, что
kolotibablo лучший из всех! Желаю дальнейшего процветания!”
81. CAPTCHA PARADOX
CAPTCHA - Completely Automated Public
Turing test to tell Computers and Humans
Apart
● CAPTCHA is a program that can
generate and grade tests that it itself
cannot pass
○ ‘much like some professors’
91. MAIN IDEA
GAN consists of a generator and a
discriminator. If the discriminator is able to
find what distinguishes fake data from real,
then the generator will learn to cheat the
discriminator.
If you make a captcha to distinguish real
data from fake, then it will be a generator
that fits on discriminator (anti-captcha)
weaknesses. The task of the captcha
generator is to fool the discriminator and,
accordingly, possible anti-captchas.