1. Introduction to Pattern Recognition Chapter 1 (Duda et al.) CS479/679 Pattern Recognition Dr. George Bebis 1

2. What is a Pattern? A pattern could be an object or event. 2 biometric patternshand gesture patterns

3. What is a Pattern? (con’t) Loan/Credit card applications – Income, # of dependents, mortgage amount for credit worthiness classification. Dating services – Age, hobbies, income for “desirability” classification. Web documents – Key-word based descriptions (e.g., documents containing “football”, “NFL”) for document classification. 3

4. Pattern Class A collection of “ similar ” objects – two challenges: – Intra-class variability – Inter-class variability 4 Letters/Numbers that look similar The letter “T” in different typefaces

5. What is Pattern Recognition? Assign a pattern to one of several known categories (or classes). 5 Gender Classification

6. What is Pattern Recognition? (cont’d) 6 Character Recognition

7. What is Pattern Recognition? (cont’d) 7 Speech Recognition

8. Modeling Pattern Classes Typically expressed in terms of a statistical model. – e.g., probability density function (Gaussian) 8 Gender Classification male female

9. Pattern Recognition Objectives Hypothesize the models that describe each pattern class (e.g., recover the process that generated the patterns). Given a novel pattern, choose the best-fitting model for it and then assign it to the pattern class associated with the model. 9

10. Classification vs Clustering – Classification (known categories) – Clustering (creation of categories) 10 Category “A” Category “B” Classification (Recognition) (Supervised Classification) Clustering (Unsupervised Classification)

11. Pattern Recognition Applications 11

12. Handwriting Recognition 12

13. License Plate Recognition 13

14. Face Detection 14 Example of unbalanced classes (i.e., faces vs non-faces)

15. Gender Classification 15 Example of balanced classes (i.e., male vs female)

16. Fingerprint Classification 16

17. Biometric Recognition 17

18. Land Cover Classification (from aerial or satellite images) 18

19. “Hot” Pattern Recognition Applications Recommendation systems – Amazon, Netflix Targeted advertising 19

20. The Netflix Prize Predict how much someone is going to enjoy a movie based on their movie preferences – $1M awarded in Sept. 2009 Can software recommend movies to customers? – Not Rambo to Woody Allen fans – Not Saw VI if you’ve seen all previous Saw movies 20

21. Main Classification Approaches Assumption: x is the input vector (pattern) y is the class label (class) Generative – Model the joint probability, p( x, y) – Make predictions by using Bayes rules to calculate p(ylx) – Pick the most likely label y Discriminative – Model p(ylx) directly, or learn a direct map from inputs x to the class labels y. – Pick the most likely label y 21

22. Syntactic Pattern Recognition Approach Represent patterns in terms of simple primitives. Describe patterns using deterministic grammars or formal languages. 22

23. Complexity of PR – An Example 23 Problem: Sorting incoming fish on a conveyor belt. Assumption: Two kind of fish: (1) sea bass (2) salmon

24. Pre-processing 24 (1) Image enhancement (2) Separating touching or occluding fish (3) Finding the boundary of each fish

25. Feature Extraction Assume a fisherman told us that a sea bass is generally longer than a salmon. We can use length as a feature and decide between sea bass and salmon according to a threshold on length. How can we choose the threshold? 25

26. “Length” Histograms Even though sea bass is longer than salmon on the average, there are many examples of fish where this observation does not hold. 26 threshold l*

27. “Average Lightness” Histograms Consider a different feature such as “average lightness” It seems easier to choose the threshold x * but we still cannot make a perfect decision. 27

28. Cost of miss-classifications There are two possible classification errors: (1) Deciding the fish was a sea bass when it was a salmon. (2) Deciding the fish was a salmon when it was a sea bass. Are both errors equally important ? 28

29. Cost of miss-classifications (cont’d) Suppose the fish packing company knows that: – Customers who buy salmon will object vigorously if they see sea bass in their cans. – Customers who buy sea bass will not be unhappy if they occasionally see some expensive salmon in their cans. How does this knowledge affect our decision? 29

30. Multiple Features To improve recognition accuracy, we might have to use more than one features at a time. – Single features might not yield the best performance. – Using combinations of features might yield better performance. 30

31. Multiple Features (cont’d) Partition the feature space into two regions by finding the decision boundary that minimizes the error. 31

32. How Many Features? Does adding more features always improve performance? – It might be difficult and computationally expensive to extract certain features. – Correlated features do not improve performance. – “Curse” of dimensionality … 32

33. Curse of Dimensionality Adding too many features can, paradoxically, lead to a worsening of performance. – Divide each of the input features into a number of intervals, so that the value of a feature can be specified approximately by saying in which interval it lies. – If each input feature is divided into M divisions, then the total number of cells is M d (d: # of features) which grows exponentially with d. – Since each cell must contain at least one point, the number of training data grows exponentially! 33

34. Model Complexity We can get perfect classification performance on the training data by choosing complex models. Complex models are tuned to the particular training samples, rather than on the characteristics of the true model. 34 How well can the model generalize to unknown samples? overfitting

35. Generalization Generalization is defined as the ability of a classifier to produce correct results on novel patterns. How can we improve generalization performance ? – More training examples (i.e., better model estimates). – Simpler models usually yield better performance. 35 complex model simpler model

36. More on model complexity 36 Regression example: Consider the following 10 sample points assuming some noise. Green curve is the true function that generated the data. Approximate the true function from the sample points.

37. More on model complexity (cont’d) 37 Polynomial curve fitting: polynomials having various orders, shown as red curves, fitted to the set of 10 sample points.

38. More on complexity (cont’d) 38 Polynomial curve fitting: 9’th order polynomials fitted to 15 and 100 sample points.

39. PR System – Two Phases 39 Training Phase Test Phase

40. PR System (cont’d) Sensing: – Use a sensor (camera or microphone) – PR depends on bandwidth, resolution, sensitivity, distortion of the sensor. Pre-processing: – Removal of noise in data. – Segmentation (i.e., isolation of patterns of interest from background). 40

41. PR System (cont’d) Training/Test data – How do we know that we have collected an adequately large and representative set of examples for training/testing the system? 41

42. PR System (cont’d) Feature extraction: – Discriminative features – Invariant features (e.g., translation, rotation and scale) – How many should we use ? – Are there ways to automatically learn which features are best ? 42

43. PR System (cont’d) Missing features: – Certain features might be missing (e.g., due to occlusion). – How should we train the classifier with missing features ? – How should the classifier make the best decision with missing features ? 43

44. PR System (cont’d) Model learning and estimation: – Models complex than necessary lead to overfitting (i.e., good performance on the training data but poor performance on novel data). – How can we adjust the complexity of the model ? (i.e., not very complex or simple). 44

45. PR System (cont’d) Classification: – Using features and learned models to assign a novel pattern to a category. – Performance can be improved using a "pool" of classifiers. – How should we build and combine multiple classifiers ? 45 "pool" of classifiers

46. PR System (cont’d) Post-processing: – Exploit context to improve performance. 46 How m ch info mation are y u mi sing?

47. PR System (cont’d) Computational Complexity: – How does an algorithm scale with the number of: features patterns categories – Consider tradeoffs between computational complexity and performance. 47

48. General Purpose PR Systems Humans have the ability to switch rapidly and seamlessly between different pattern recognition tasks. It is very difficult to design a system that is capable of performing a variety of classification tasks. – Different decision tasks may require different features. – Different features might yield different solutions. – Different tradeoffs exist for different tasks. 48