MACHINE LEARNING 1. Introduction

What is Machine Learning? Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 2  Need an algorithm to solve a problem on a computer  An Algorithm is a sequence of instructions to transform input from output  Example: Sort list of numbers  Input: set of numbers  Output: ordered list of numbers  Many algorithms for the same task  May be interested in finding the most efficient

What is Machine Learning? Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 3  Don’t have an algorithm for some tasks  Example: Tell the spam for legitimate s  Know the input (an ) and output (yes/no)  Don’t know how to transform input to output  Definition of spam may change over the time and from individual to individual  We don’t have a knowledge, replace it with data  Can easily produce large amount of examples  Want a computer to extract an algorithm from the examples

What is Machine Learning? Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 4  Believe that there is a process explaining the data  We don’t know details about the process we know it’s not random  Example: Consumer Behavior  Frequently buy beer with chips  Buy more ice-cream ins summer  There is certain patterns in data  Rarely can’t indentify patterns completely  Can construct good and useful approximation

Approximations to Patterns Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 5  May not explain everything  Still detect some patterns and regularities  Use patterns  Understand the process  Make a prediction

Data Mining: Application of ML to large databases Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 6  Retail: Market basket analysis, Customer relationship management (CRM)  Finance: Credit scoring, fraud detection  Manufacturing: Optimization, troubleshooting  Medicine: Medical diagnosis  Telecommunications: Quality of service optimization  Bioinformatics: Motifs, alignment  Web mining: Search engines ...

Examples of Ml Applications Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 7  Learning association: Basket Analysis  If people buy X they typically buy Y  There is a customer who buys X and don’t buy Y  He/She is a potential Y customer  Find such customers and target them for cross-selling  Find an association rule: P(Y|X)  D customer attributes (e. g.age, gender)  P(Y|X,D)

Association Rules examples Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 8  Bookseller instead of supermarket  Products are books or authors  Web portal  Links the user is likely to click  Pre-download pages in advance

Credit Scoring Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 9  Bank want to predict a risk associated with loan  Probability that customer will default given the information about the customer  Income, Savings, profession  Association between customer attributes and his risk  Fits a model to the past data to be able to calculate a risk for a new application  Accept/Refuse application

Classification Problem Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 10  Two classes of customers: low-risk and high-risk  Input: information about a customer  Output: assignment to one of two classes  Example of classification rule  IF income> θ 1 AND savings> θ 2 THEN low-risk ELSE high-risk  Discriminant: function that separates the examples of different classes

Discriminant Rule Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 11

Discriminant Rule Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 12  Prediction: User rule for novel instances  In some instances may want to calculate probabilities instead of 0/1  P(Y|X), P(Y=1|X=x) =0.8, customer has an 80% probability of being high-risk

Pattern Recognition Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 13  Optical character recognition (OCR), recognizing character codes from their images  Number of classes as many as number of images we would like to recognize  Handwritten characters (zip code on envelopes or amounts on checks)  Different handwriting styles, character sizes, pen or pencil  We don’t have a formal description that covers all A’s characters and none of non-A’s

OCR Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 14  All A have something in common  Extract pattern from examples  Use redundancy in human languages  Word is a sequence of characters  Not all sequences are equally likely  Can still r?ad some w?rds  ML algorithms should model dependencies among characters in a word and word in the sequence

Face recognition Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 15  Input : image  Classes: people to be recognized  Learning program: learn to associate faces to identities  More difficult then OCR  More classes  Images are larger  Differences in pose and lightening cause significant changes in image  Occlusions: Glasses, beard

Face Recognition Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 16 Training examples of a person Test images AT&T Laboratories, Cambridge UK

Medical Diagnosis Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 17  Input: Information about the patient  Age, past medical history, symptoms  Output: illnesses  Can apply some additional tests  Costly and inconvenient  Some information might be missing  Can decide to apply test if believe valuable  High price of wrong decision

Speech Recognition Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 18  Input: acoustic signal  Output: words  Different accents and voices  Can integrate language models  Combine with lips movement  Sensor fusion

Knowledge Extraction Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 19  The rule is simpler than the data  Example: Discriminant separating low-risk and high risk customer helps to define low risk customer  Target low risk customer through advertising

Compression Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 20  Explanation simpler than data  Discard the data, keep the rule  Less memory  Example: Image compression  Learn most common colors in image  Represent slightly different but similar colors by single value

Outlier detection Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 21  Find instances which do not obey rules  Interesting not in a rule but an exception not covered by the rule  Examples: Learn properties of standard credit card transactions  Outlier is a suspected fraud

Why “Learn” ? Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 22  Machine learning is programming computers to optimize a performance criterion using example data or past experience.  There is no need to “learn” to calculate payroll  Learning is used when:  Human expertise does not exist (navigating on Mars),  Humans are unable to explain their expertise (speech recognition)  Solution changes in time (routing on a computer network)  Solution needs to be adapted to particular cases (user biometrics)

What We Talk About When We Talk About“Learning” Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 23  Learning general models from a data of particular examples  Data is cheap and abundant (data warehouses, data marts); knowledge is expensive and scarce.  Example in retail: Customer transactions to consumer behavior: People who bought “Da Vinci Code” also bought “The Five People You Meet in Heaven” (  Build a model that is a good and useful approximation to the data.

What is Machine Learning? Based On E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 24  Optimize a performance criterion using example data or past experience.  Role of Statistics: Inference from a sample  Role of Computer science: Efficient algorithms to  Solve the optimization problem  Representing and evaluating the model for inference

Regression  Example: Price of a used car  x : car attributes y : price y = g (x | θ  ) g ( ) model, θ parameters Based on E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 25 y = wx+w 0

Supervised Learning Lecture Notes for E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 26  Regression and classification are supervised learning problems  There is an input and output  Need to learn mapping from input to output  ML approach: assume a model defined up to a set of parameters  y = g(x| θ )  Machine Learning Program optimize the parameters to minimize the error  Linear model might be two restrictive (large error)  Use more complex models  y = w 2 x 2 + w 1 x + w 0

Supervised Learning: Uses Lecture Notes for E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 27  Prediction of future cases: Use the rule to predict the output for future inputs  Knowledge extraction: The rule is easy to understand  Compression: The rule is simpler than the data it explains  Outlier detection: Exceptions that are not covered by the rule, e.g., fraud

Unsupervised Learning Based On E Alpaydın 2004 Introduction to Machine Learning © The MIT Press (V1.1) 28  Learning “what normally happens”  No output, only input  Statistics: Density estimation  Clustering: Grouping similar instances  Example applications  Customer segmentation in CRM  Image compression: Color quantization  Document Clustering

Reinforcement Learning Based on Introduction to Machine Learning © The MIT Press (V1.1) 29  Learning a policy: A sequence of outputs  Single action is not important  Action is good if its part of a good policy  Learn from past good policies  Delayed reward  Example: Game playing  Robot in a maze  Reach the goal state from an initial state