Machine learning methods such as neural networks and self-organizing maps can be used to perform CAPTCHA recognition with high accuracy. Five experiments were conducted comparing these methods on CAPTCHAs of varying length and character sets. The results showed that accuracy decreases as CAPTCHAs increase in length due to worsening segmentation quality. Better segmentation techniques are needed to maintain high true positive rates. Future work involves improving segmentation and ensemble methods to boost recognition rates on longer CAPTCHAs.

1. Machine Learning Methods
for CAPTCHA Recognition
Rachel Shadoan
Zachery Tidwell, II

2. CAPTCHA
Completely Automated Public Turing Test to tell Computers and Humans Apart
Why are they interesting?
o Harder than normal text recognition
On par with handwriting recognition,
reading damaged text
o Techniques translate well to other problems
Facial recognition (Gonzaga, 2002)
Weed identification (Yang, 2000)
o Near infinite data sets
Easier to avoid over-fitting

3. Hypothesis
CAPTCHA recognition can be
accomplished to a high degree
of accuracy using machine
learning methods with minimal
preprocessing of inputs.

4. Methods
Tools
o JCaptcha
o Image Processing
Learning Methods Segmentation Methods
o Feed-forward Neural o Overlapping
Nets o Whitespace
o Self-Organizing Maps o K-Means
o K-Means
o Cluster Classification

5. JCaptcha
o Open-source CAPTCHA
generation software
o Highly configurable
Can produce CAPTCHAs of
many levels of difficulty
o Check it out at:
http://jcaptcha.sourceforge.net

6. Image Processing
Sparse Image
Represents Images as unbounded set of pixels
Each pixel is a value between 0 and 1 and a
coordinate pair
Center each image before turning into a matrix of
0s and 1s
Original After Transformation

7. Feed-Forward Neural Nets
As covered in class

8. Self-Organizing Maps
Training Collection
Initialize N buckets to For many inputs
random values
Sort each input into
For each input the bucket it most
Find the bucket that is closely matches
“closest” to the input For each bucket and each
Adjust the “closest” character
bucket to more closely Calculate the
match the input using probability of that
exponential average character going into
that bucket.

9. K-Means
• Very similar to Self‐
Organizing Maps
(SOMs)
• Can use the same
classifying mechanism
as used for SOM

10. Overlapping Segmentation
• Divide image into
fixed number of
overlapping tiles of
the same size
• In our case, 20 x 20
pixels with a 50%
overlap
• Discard chunks
under a certain size Note: This is a B with
part of it cut off, not
and chunks that are an E. Therein lies the
all white rub.

11. Whitespace Segmentation
• Iterate through the
image from left to
right—segment
when a full column
of whitespace is
encountered
• Works perfectly for
well-spaced text

12. K-Means Segmentation
• Performs better
than heuristic
segmentation on
closely-packed
inputs

13. Segmentation Comparison
Even‐width
Whitespace
K‐Means
Even‐width
Whitespace
K‐Means

14. Experiment 1
Machine Learning Method:
Self-Organizing Map
Topology
200 buckets, initialized randomly
Inputs:
3 letter CATPCHAs
Random fonts
Letters A-G
“Chunked” using overlapping segmentation

15. Experiment 1 Results
Buckets fell into three primary categories:
Distinguishable
letters
Chunks with halves
of two letters
Indistinguishable
noise

16. Experiment 1 Results

17. Experiment 2
ML Method: Contains … ?
Neural Net
A: 0 or 1
Topology: B : 0 or 1
C: 0 or 1
400 Nodes
Fully connected
50 Nodes
7 Nodes
D: 0 or 1
E: 0 or 1
400 inputs F: 0 or 1
50 node hidden layer G: 0 or 1
7 outputs
Inputs:
Single letter CATPCHAs
Random fonts
Letters A-G

18. Experiment 2 Results
Neural Net Learning Curve

19. Experiment 2 Results
Past a certain
number of nodes
in the hidden
layer, the
topology ceases
to have a huge
impact on
accuracy.
Neural Net Accuracy vs. Size of Hidden Layer

20. Experiment 3
ML Method: ML Method:
SOM Neural Net
Topology: Topology:
500 buckets Fully connected
400 inputs
1000 node hidden layer
7 outputs
Inputs:
4 letter CATPCHAs
Fandom fonts
Letters A-G

21. Experiment 3
Neural Net vs. SOM on CAPTCHAs Length 4, Letters A‐G

22. Experiment 4
ML Method: ML Method:
SOM Neural Net
Topology: Topology:
500 buckets Fully connected
400 inputs
1000 node hidden layer
7 outputs
Inputs:
4 letter CATPCHAs
Fandom fonts
Letters A-Z

23. Experiment 4
Neural Net vs. SOM on CAPTCHAs Length 4, Letters A‐Z

24. Experiment 5
ML Method: ML Method:
SOM Neural Net
Topology: Topology:
500 buckets Fully connected
400 inputs
1000 node hidden layer
7 outputs
Inputs:
5 letter CATPCHAs
Fandom fonts
Letters A-Z

25. Experiment 5
Neural Net vs. SOM on CAPTCHAs Length 5, Letters A-Z

26. What it all means
• Increasing number of characters
dramatically decreases total accuracy
because segmentation quality decreases
• True positive rate goes down when
segmentation quality decreases
• Hence, better segmentation is the key

27. Future Work
Improved Segmentation
o Wirescreen segmentation
o Ensemble techniques
Improved True Positive Rates with Current
System
o Ensemble techniques
New problems
o Handwriting recognition
o Bot net of doom

28. Questions?