Classification

Classification

•          Select the length of the fish as a possible feature for discrimination

 

•          The length is a poor feature alone!

•          Select the lightness as a possible feature.

 

•          Threshold decision boundary and cost relationship

Move our decision boundary toward smaller values of lightness in order to minimize the cost (reduce the number of sea bass that are classified salmon!)

 

•          Adopt the lightness and add the width of the fish

 

 

•          We might add other features that are not correlated with the ones we already have. A precaution should be taken not to reduce the performance by adding such “noisy features”

Ideally, the best decision boundary should be the one which provides an optimal performance such as in the following figure

 

However, our satisfaction is premature because the central aim of designing a classifier is to correctly classify novel input   

 

•          Sensing

•          Use of a transducer (camera or microphone)

•          PR system depends of the bandwidth, the resolution sensitivity distortion of the transducer

•          Segmentation and grouping

•          Patterns should be well separated and should not overlap

 

•          Feature extraction

•          Discriminative features

•          Invariant features with respect to translation, rotation and scale.

•          Classification

•          Use a feature vector provided by a feature extractor to assign the object to a category

•          Post Processing

•          Exploit context -input dependent information other than from the target pattern itself- to improve performance