Classification Examples
Classification Examples
Classification Examples
Classification Examples
 Classification has many applications in real life.
 From last lecture:
 Weather Prediction
 Image Classification
 Document Classification
 More examples next !
Face Recognition
Face Recognition
 Input : an image
 Output: Specifying the location of the face (if exists)
 More advanced: A specific person face recognition.
Road Sign Recognition
Road Sign Recognition
 Input : an image of a road sign (taken by a driverless car)
 Output: recognizing the sign
Tumor Recognition
Tumor Recognition
 Input : an MRI image of the patient.
 Output: recognizing if it has tumor or not.
Handwriting Recognition
Handwriting Recognition
 Input : an image that contains handwriting.
 Output: recognizing the handwriting.
Medical Diagnosis
Medical Diagnosis
Training
Testing
Temp Tired? Cough? Smell … Corona?
27 Y Y N Y
39 Y Y Y Y
27 N N Y N
31 Y N N ??
Sentiment Analysis
Sentiment Analysis
 Classifying people feelings toward certain product,
decision, topic, …
Training
“ This restaurant is awesome “ , label = positive
“ I do not like this car” , label = negative
…
Testing
“ This phone is just amazing “ , label = ?
 Can be used by companies to evaluate and improve their
products using the reviews.
Classifiers
Classifiers
 A classifier is a method that learns from training data to predict
the labels of testing data.
 There are many classifiers, each classifier has its own method.
 Different classifiers have different accuracies.
Classifiers
Classifiers
 Examples of classifiers:
 K Nearest Neighbor (KNN)
 Decision Trees
 Logistic Regression
 Artifical Neural Networks (ANN)
 Support Vector Machines (SVM)
 Rule-Based Classifiers
 Bayesian Classifiers
Selection of Classifiers
Selection of Classifiers
 There are many classifiers out there, which classifier should I
use ?
 It depends on the problem.
 Each classifier has its own advantages and disadvantages.
 For fast training time, we use Naïve Bayes or KNN, we do not
use decision tree or ANN
 For high accuracy, we use ANN or SVM
 For fast prediction, we use decision tree, not KNN
 Dataset is small, do not use ANN
 Dataset is very large, use ANN, do not use SVM

2.Classification in Artificial Neural Network Examples.ppt

  • 1.
  • 2.
    Classification Examples Classification Examples Classification has many applications in real life.  From last lecture:  Weather Prediction  Image Classification  Document Classification  More examples next !
  • 3.
    Face Recognition Face Recognition Input : an image  Output: Specifying the location of the face (if exists)  More advanced: A specific person face recognition.
  • 4.
    Road Sign Recognition RoadSign Recognition  Input : an image of a road sign (taken by a driverless car)  Output: recognizing the sign
  • 5.
    Tumor Recognition Tumor Recognition Input : an MRI image of the patient.  Output: recognizing if it has tumor or not.
  • 6.
    Handwriting Recognition Handwriting Recognition Input : an image that contains handwriting.  Output: recognizing the handwriting.
  • 7.
    Medical Diagnosis Medical Diagnosis Training Testing TempTired? Cough? Smell … Corona? 27 Y Y N Y 39 Y Y Y Y 27 N N Y N 31 Y N N ??
  • 8.
    Sentiment Analysis Sentiment Analysis Classifying people feelings toward certain product, decision, topic, … Training “ This restaurant is awesome “ , label = positive “ I do not like this car” , label = negative … Testing “ This phone is just amazing “ , label = ?  Can be used by companies to evaluate and improve their products using the reviews.
  • 9.
    Classifiers Classifiers  A classifieris a method that learns from training data to predict the labels of testing data.  There are many classifiers, each classifier has its own method.  Different classifiers have different accuracies.
  • 10.
    Classifiers Classifiers  Examples ofclassifiers:  K Nearest Neighbor (KNN)  Decision Trees  Logistic Regression  Artifical Neural Networks (ANN)  Support Vector Machines (SVM)  Rule-Based Classifiers  Bayesian Classifiers
  • 11.
    Selection of Classifiers Selectionof Classifiers  There are many classifiers out there, which classifier should I use ?  It depends on the problem.  Each classifier has its own advantages and disadvantages.  For fast training time, we use Naïve Bayes or KNN, we do not use decision tree or ANN  For high accuracy, we use ANN or SVM  For fast prediction, we use decision tree, not KNN  Dataset is small, do not use ANN  Dataset is very large, use ANN, do not use SVM