Lecture 6_LDA2 - Classification: Supervised learning, and...

Info iconThis preview shows pages 1–15. Sign up to view the full content.

View Full Document Right Arrow Icon
4/14/11 Classification: Supervised learning, and Model Evaluation Introduction to Discriminant Analysis and Classification Tommy W. S. Chow 11
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
4/14/11 What is classification 22
Background image of page 2
4/14/11 Illustrating Classification Task i.e., Classify if a man is a business man or not 33
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
4/14/11 Classification: Definition l Given a collection of records ( training set ) - Each record contains a set of attributes , one of the attributes is the class . l Find a model for class attribute as a function of the values of other attributes. l Goal: previously unseen records should be assigned a class as accurately as possible. - A test set is used to determine the accuracy of the model. Usually, the given data set is divided into training and test sets, with training set used to build the model and test set used to validate it. 44
Background image of page 4
4/14/11 Examples of Classification Task   l Predicting tumor cells as benign or malignant l Classifying credit card transactions as legitimate or fraudulent l Classifying faulty or normal system operations. If it is a faulty system, what type of fault is it? l Categorizing news stories, finance, politics, sport, entertainment, etc l Classifying problematic students from good student l Predicting US president election, McCain or Obama Brazil Black Out 10 Nov 2009 55
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
4/14/11 Fish classification: Sea bass or salmon Histograms for the lightness feature. It shows no single threshold value x* (decision boundary) will be prefect for separating the two classes completely 66
Background image of page 6
4/14/11 Using two features together (2-dimensional case) Using the 2 features together. The dark line is the decision boundary. Overall classification is higher than using any of the 1 feature alone Too complex model for the fish leads to too complicated model. It has high classification rate in this data set but will surely lead to poor performance for future patterns An optimal decision boundary tradeoff between performance on the training set and future classification data 77
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
4/14/11 Classification Techniques Generative model This approach is based on statistical and probabilistic modeling of the happenings: such as Linear Discriminate Analysis (LDA), and Naïve Bayes classifier Discriminative model l Decision Tree based Methods l Support Vector Machine (SVM) l Neural Networks l K Nearest Neighbor 88
Background image of page 8
4/14/11 48,000 records, 16 attributes [Kohavi 1995] 99
Background image of page 9

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
4/14/11 Predicting wealth from age Carnegie Mellon University (CMU) , AI notes 1010
Background image of page 10
4/14/11 Predicting wealth from age Persons whose age are 20-30 are the most possible poor; ones whose age are 50-60 are the most possible rich Carnegie Mellon University (CMU) , AI notes 1111
Background image of page 11

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
4/14/11 Wealth from hours worked Carnegie Mellon University (CMU) , AI notes 1212
Background image of page 12
4/14/11 Wealth from years of education Persons with higher education are the most possible rich Carnegie Mellon University (CMU) , AI notes 1313
Background image of page 13

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
4/14/11 Discriminat Analysis Introduction l Discriminant Analysis is a classic method of classification that often produces models whose accuracy approaches more complex modern methods.
Background image of page 14
Image of page 15
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 04/14/2011 for the course EE 4146 taught by Professor Tommychow during the Spring '11 term at City University of Hong Kong.

  • Spring '11
  • TommyChow