Pattern Recognition

10,330 views
10,267 views

Published on

0 Comments
4 Likes
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total views
10,330
On SlideShare
0
From Embeds
0
Number of Embeds
46
Actions
Shares
0
Downloads
987
Comments
0
Likes
4
Embeds 0
No embeds

No notes for slide

Pattern Recognition

  1. 1. PATTERN RECOGNITION Talal A. Alsubaie SFDA
  2. 2. OUTLINES <ul><li>What is a pattern? </li></ul><ul><li>What is A pattern Class ? </li></ul><ul><li>What is pattern recognition? </li></ul><ul><li>Human Perception </li></ul><ul><li>Examples of applications </li></ul><ul><li>The Statistical Way </li></ul><ul><li>Human and Machine Perception </li></ul><ul><li>P attern R ecognition </li></ul><ul><li>Pattern Recognition Process </li></ul><ul><li>Case Study </li></ul>
  3. 3. WHAT IS A PATTERN? <ul><li>A pattern is an abstract object, or a set of measurements describing a physical object. </li></ul>
  4. 4. WHAT IS A PATTERN CLASS ? <ul><li>A pattern class (or category) is a set of patterns sharing common attributes. </li></ul><ul><li>A collection of “similar” (not necessarily identical) objects. </li></ul><ul><li>During recognition given objects are assigned to prescribed classes. </li></ul>
  5. 5. WHAT IS PATTERN RECOGNITION? <ul><li>Theory, Algorithms, Systems to put Patterns into Categories </li></ul><ul><li>Relate Perceived Pattern to Previously Perceived Patterns </li></ul><ul><li>Learn to distinguish patterns of interest from their background </li></ul>
  6. 6. HUMAN PERCEPTION <ul><li>Humans have developed highly sophisticated skills for sensing their environment and taking actions according to what they observe, e.g., </li></ul><ul><ul><li>Recognizing a face. </li></ul></ul><ul><ul><li>Understanding spoken words. </li></ul></ul><ul><ul><li>Reading handwriting. </li></ul></ul><ul><ul><li>Distinguishing fresh food from its smell. </li></ul></ul><ul><li>We would like to give similar capabilities to machines. </li></ul>
  7. 7. EXAMPLES OF APPLICATIONS
  8. 8. THE STATISTICAL WAY
  9. 9. GRID BY GRID COMPARISON Grid by Grid Comparison A A B
  10. 10. GRID BY GRID COMPARISON 0 0 1 0 0 0 1 0 0 1 1 1 1 0 0 1 1 0 0 1 No of Mismatch= 3 A A B 0 1 1 0 0 1 1 0 0 1 1 0 1 0 0 1 1 0 0 1
  11. 11. GRID BY GRID COMPARISON Grid by Grid Comparison A A B
  12. 12. GRID BY GRID COMPARISON 0 0 1 0 0 0 1 0 0 1 1 1 1 0 0 1 1 0 0 1 1 1 1 0 0 1 0 1 0 1 1 1 0 1 0 1 1 1 1 0 No of Mismatch= 9 A A B
  13. 13. PROBLEM WITH GRID BY GRID COMPARISON <ul><li>Time to recognize a pattern - Proportional to the number of stored patterns ( Too costly with the increase of number of patterns stored ) </li></ul>Solution Artificial Intelligence A-Z a-z 0-9 */-+1@#
  14. 14. HUMAN AND MACHINE PERCEPTION <ul><li>We are often influenced by the knowledge of how patterns are modeled and recognized in nature when we develop pattern recognition algorithms. </li></ul><ul><li>Research on machine perception also helps us gain deeper understanding and appreciation for pattern recognition systems in nature. </li></ul><ul><li>Yet, we also apply many techniques that are purely numerical and do not have any correspondence in natural systems. </li></ul>
  15. 15. PATTERN RECOGNITION <ul><li>Two Phase : Learning and Detection . </li></ul><ul><li>Time to learn is higher. </li></ul><ul><ul><li>Driving a car </li></ul></ul><ul><li>Difficult to learn but once learnt it becomes natural . </li></ul><ul><li>Can use AI learning methodologies such as: </li></ul><ul><ul><li>Neural Network. </li></ul></ul><ul><ul><li>Machine Learning. </li></ul></ul>
  16. 16. LEARNING <ul><li>How can machine learn the rule from data? </li></ul><ul><ul><li>Supervised learning : a teacher provides a category label or cost for each pattern in the training set. </li></ul></ul><ul><ul><li>Unsupervised learning : the system forms clusters or natural groupings of the input patterns. </li></ul></ul>
  17. 17. <ul><li>Classification (known categories) </li></ul><ul><li>Clustering (creation of new categories) </li></ul>CLASSIFICATION VS. CLUSTERING Category “A” Category “B” Clustering (Unsupervised Classification) Classification (Supervised Classification)
  18. 18. PATTERN RECOGNITION PROCESS (CONT.) Post- processing Classification Feature Extraction Segmentation Sensing input Decision
  19. 19. PATTERN RECOGNITION PROCESS <ul><li>Data acquisition and sensing: </li></ul><ul><ul><li>Measurements of physical variables. </li></ul></ul><ul><ul><li>Important issues: bandwidth, resolution , etc. </li></ul></ul><ul><li>Pre-processing: </li></ul><ul><ul><li>Removal of noise in data. </li></ul></ul><ul><ul><li>Isolation of patterns of interest from the background. </li></ul></ul><ul><li>Feature extraction: </li></ul><ul><ul><li>Finding a new representation in terms of features. </li></ul></ul><ul><li>Classification </li></ul><ul><ul><li>Using features and learned models to assign a pattern to a category. </li></ul></ul><ul><li>Post-processing </li></ul><ul><ul><li>Evaluation of confidence in decisions. </li></ul></ul>
  20. 20. CASE STUDY <ul><li>Fish Classification: </li></ul><ul><ul><li>Sea Bass / Salmon. </li></ul></ul><ul><li>Problem : Sorting incoming fish </li></ul><ul><li>on a conveyor belt according to </li></ul><ul><li>species. </li></ul><ul><li>Assume that we have only two kinds of fish: </li></ul><ul><ul><li>Sea bass. </li></ul></ul><ul><ul><li>Salmon. </li></ul></ul>Salmon Sea-bass
  21. 21. CASE STUDY (CONT.) <ul><li>What can cause problems during sensing? </li></ul><ul><ul><li>Lighting conditions. </li></ul></ul><ul><ul><li>Position of fish on the conveyor belt. </li></ul></ul><ul><ul><li>Camera noise. </li></ul></ul><ul><ul><li>etc… </li></ul></ul><ul><li>What are the steps in the process? </li></ul><ul><ul><li>Capture image. </li></ul></ul><ul><ul><li>Isolate fish </li></ul></ul><ul><ul><li>Take measurements </li></ul></ul><ul><ul><li>Make decision </li></ul></ul>
  22. 22. CASE STUDY (CONT.) Classification Feature Extraction Pre-processing “ Sea Bass” “ Salmon”
  23. 23. CASE STUDY (CONT.) <ul><li>Pre-Processing: </li></ul><ul><ul><li>Image enhancement </li></ul></ul><ul><ul><li>Separating touching or occluding fish. </li></ul></ul><ul><ul><li>Finding the boundary of the fish. </li></ul></ul>
  24. 24. HOW TO SEPARATE SEA BASS FROM SALMON? <ul><li>Possible features to be used: </li></ul><ul><ul><li>Length </li></ul></ul><ul><ul><li>Lightness </li></ul></ul><ul><ul><li>Width </li></ul></ul><ul><ul><li>Number and shape of fins </li></ul></ul><ul><ul><li>Position of the mouth </li></ul></ul><ul><ul><li>Etc … </li></ul></ul><ul><li>Assume a fisherman told us that a “sea bass” is generally longer than a “salmon”. </li></ul><ul><li>Even though “sea bass” is longer than “salmon” on the average, there are many examples of fish where this observation does not hold. </li></ul>
  25. 25. HOW TO SEPARATE SEA BASS FROM SALMON? <ul><li>To improve recognition, we might have to use more than one feature at a time. </li></ul><ul><ul><li>Single features might not yield the best performance. </li></ul></ul><ul><ul><li>Combinations of features might yield better performance. </li></ul></ul>
  26. 26. FEATURE SELECTION “ Good” features “ Bad” features
  27. 27. DECISION BOUNDARY
  28. 28. DECISION BOUNDARY (CONT.) More complex model result more complex boundary
  29. 29. DECISION BOUNDARY (CONT.) Different criteria lead to different decision boundaries
  30. 30. DECISION BOUNDARY (CONT.) <ul><li>What if a customers find “ Sea bass ” in there “ Salmon ” can? </li></ul><ul><li>We should also consider costs of different errors we make in our decisions. </li></ul>
  31. 31. DECISION BOUNDARY (CONT.) <ul><li>For example, if the fish packing company knows that: </li></ul><ul><ul><li>Customers who buy salmon will object vigorously if they see sea bass in their cans. </li></ul></ul><ul><ul><li>Customers who buy sea bass will not be unhappy if they occasionally see some expensive salmon in their cans. </li></ul></ul><ul><li>How does this knowledge affect our decision? </li></ul>
  32. 32. CASE STUDY (CONT.) <ul><li>Issues with feature extraction: </li></ul><ul><ul><li>Correlated features do not necessary improve performance. </li></ul></ul><ul><ul><li>It might be difficult to extract certain features. </li></ul></ul><ul><ul><li>It might be computationally expensive to extract many features. </li></ul></ul><ul><ul><li>Missing Features. </li></ul></ul><ul><ul><li>Domain Knowledge. </li></ul></ul>
  33. 33. THE DESIGN CYCLE
  34. 34. DEMO
  35. 35. DEMO <ul><li>Online face detector demo: </li></ul><ul><ul><li>http://demo.pittpatt.com/index.php </li></ul></ul>
  36. 36.
  37. 37. DEMO (CONT.) <ul><li>With my friend “ Albert Einstein ” </li></ul>
  38. 38. VIDEO DEMO
  39. 39. Q & A
  40. 40. THANK YOU

×