This preview shows pages 1–5. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: Insight by Mathematics and Intuition for understanding Pattern Recognition Waleed A. Yousef Faculty of Computers and Information, Helwan University. May 8, 2010 Ch4. Linear Models for Classification Before modeling a linear model for classification, what is the best descision function? Back to Ch2. we find: ln f 1 ( X ) f 2 ( X ) G 1 ≷ G 2 ln π 2 L 21 π 1 L 12 h ( X ) G 1 ≷ G 2 th, (the loglikelihood ratio). If we know f 1 and f 2 the best thing one can do is to use them and estimate their parameters, and h ( X ) is the decision function that decides to which class X belongs, with the decision surface h ( X ) = th . Decision boundary for mltinormal distributions 1 2 x Í Σ 1 2 Σ 1 1 x ü ûú ý Quadratic Term x Í Σ 1 2 μ 2 Σ 1 1 μ 1 ü ûú ý Linear Term + 1 2 μ Í 2 Σ 1 2 μ 2 μ Í 1 Σ 1 1 μ 1 1 2 ln  Σ 1   Σ 2  = th Linear Discriminant Analysis (LDA): We assume that...
View
Full
Document
 Spring '10
 WaleedA.Yousef

Click to edit the document details