CAPTCHA and Convolutional neural network

22
CAPTCHA

Transcript of CAPTCHA and Convolutional neural network

Page 1: CAPTCHA and Convolutional neural network

CAPTCHA

Page 2: CAPTCHA and Convolutional neural network

CAPTCHA •A CAPTCHA (an acronym for "Completely

Automated Public Turing test to tell Computers and Humans Apart")

Is a program that protects websites against bots by generating and grading tests that human can pass but current computers cannot .

Page 3: CAPTCHA and Convolutional neural network

Applications of CAPTCHAs

-Preventing comment spam in blogs

-Protecting websites registration

-Protecting email address from scrapers

-Online polls

-Preventing dictionary attacks

-Spam and worms

Page 4: CAPTCHA and Convolutional neural network

Types of CAPTCHAs

The most common is :

Visual CAPTCHA

Series of characters

and digits

Page 5: CAPTCHA and Convolutional neural network

Design of visual CAPTCHAs

For more difficulty and security we add:

-Text : color , font size , font type , style , set used

-Background : color , texture

-Translation , Rotation , Scaling ,over lapping , clutter , crossing lines , distortion , waving .

3D CAPTCHAs .

But Be careful

Page 6: CAPTCHA and Convolutional neural network

Design of visual CAPTCHAs

Page 7: CAPTCHA and Convolutional neural network

Breaking CAPTCHAs

1- Preprocessing .

2- Segmentation .

Page 8: CAPTCHA and Convolutional neural network

Breaking CAPTCHAs

3-Post segmentation (Normalizing) .

4- Recognition .

5- Post recognition .

Final Output is : “pvack”

Page 9: CAPTCHA and Convolutional neural network

Some CAPTCHA defeating projects

-Mori et all

-PWNTCHA from

-Microsoft research

-Vicarious (AI company)

Page 10: CAPTCHA and Convolutional neural network

Microsoft research results

Page 11: CAPTCHA and Convolutional neural network

Microsoft research results

Page 12: CAPTCHA and Convolutional neural network

Microsoft research results

Page 13: CAPTCHA and Convolutional neural network

Microsoft research results

Page 14: CAPTCHA and Convolutional neural network

So … where is the neural network

??

Convolutional Neural Network

Page 15: CAPTCHA and Convolutional neural network

Convolutional neural network

-Type of feed-forward MLP.

-Conv. Nets are inspired by biological processes in visual cortex.

-So it is used in image recognition and handwritten recognition .

-high performance in MNIST database .

-Designed by Yann Lecun .

Page 16: CAPTCHA and Convolutional neural network

Convolution

-Convolution is a common image processing technique that changes the intensities of a pixel to reflect the intensities of the surrounding pixels. A common use of convolution is to create image filters

Page 17: CAPTCHA and Convolutional neural network

Sparse Connectivity

Page 18: CAPTCHA and Convolutional neural network

Types of layers

-Convolutional layers .

-Kernel

-Feature Map or filter .

-Shared weights .

-Subsampling or pooling .

-Full connected layer (classification) .

Page 19: CAPTCHA and Convolutional neural network

Le Net

Page 20: CAPTCHA and Convolutional neural network

Modified Back propagation

-only a small change to the original algorithm. The gradient of a shared weight is simply the sum of the gradients of the parameters being shared.

w1=w2 ∆w1=∆w2

𝜕𝐸

𝜕𝑤2

𝜕𝐸

𝜕𝑤1

Sum

Page 21: CAPTCHA and Convolutional neural network

Solution !!!

-Image recognition Deep learning

-3D CAPTCHA.

-moving image .

-Audio CAPTCHA .

-MAPTCA (Mathematical CAPTCHA).

But…..

Page 22: CAPTCHA and Convolutional neural network

Conclusion

But …

It is trivial for AI.

“Any program that passes the tests generated by a CAPTCHA can be used to solve a hard unsolved AI problem”

Here we use knowledge for evil