SlideShare a Scribd company logo
1 of 61
Machine Learning Version Spaces Learning
Introduction: ,[object Object],[object Object],[object Object],[object Object],[object Object],[object Object],[object Object],[object Object],[object Object],[object Object]
Version Spaces A  concept learning  technique based on  refining models  of the world.
Concept Learning ,[object Object],[object Object],Restaurant Meal Day Cost    Reaction Alma 3 breakfast Friday cheap Yes De Moete lunch Friday expensive No Alma 3 lunch Saturday cheap Yes Sedes breakfast Sunday cheap No Alma 3 breakfast Sunday expensive No + - + - - ,[object Object]
In general ,[object Object],[object Object],[object Object],[object Object],[object Object],Restaurant Meal Day Cost    3  X  3  X  7  X  2  =  126 Reaction:   Restaurant X Meal X Day X Cost  -->  Bool Find an inductive ‘guess’ of the concept, that covers all the examples !
Pictured: Set of all possible events                                   Given some  examples: + + + + - - - - Find a concept that covers all the positive  examples and none of the negative ones!
Non-determinism ,[object Object],[object Object],Set of all possible events                                   + + + + - - - -
An obvious, but bad choice: ,[object Object],[object Object],[object Object],[object Object],Restaurant Meal Day Cost    Reaction Alma 3 breakfast Friday cheap Yes De Moete lunch Friday expensive No Alma 3 lunch Saturday cheap Yes Sedes breakfast Sunday cheap No Alma 3 breakfast Sunday expensive No - - - + +
Pictured: ,[object Object],Set of all possible events + + + + - - - -                                  
Equally bad is: ,[object Object],[object Object],[object Object],[object Object],Restaurant Meal Day Cost    Reaction Alma 3 breakfast Friday cheap Yes De Moete lunch Friday expensive No Alma 3 lunch Saturday cheap Yes Sedes breakfast Sunday cheap No Alma 3 breakfast Sunday expensive No + - + - -
Pictured: ,[object Object],Set of all possible events                                   + + + + - - - -
Solution:  fix a language of hypotheses: ,[object Object],= hypothesis space ,[object Object],[object Object],[object Object]
Reaction - Example: Restaurant Meal Day Cost    Reaction Alma 3 breakfast Friday cheap Yes De Moete lunch Friday expensive No Alma 3 lunch Saturday cheap Yes Sedes breakfast Sunday cheap No Alma 3 breakfast Sunday expensive No + - + - - ,[object Object],[object Object],[object Object],[object Object],[object Object],[object Object]
Hypotheses relate to  sets of possible events        Events    x2      x1 x1 =  <  Alma 3, lunch, Monday, expensive> x2 = < Sedes, lunch, Sunday, cheap> h1  = [?, lunch, Monday, ?] h2  = [?, lunch, ?, cheap] h3  = [?, lunch, ?, ?] General Specific                 Hypotheses h2 h3 h1
Expressive power of this hypothesis language: ,[object Object],[object Object],[object Object],[object Object],[object Object],[object Object],[object Object],[object Object]
Other languages of hypotheses are allowed ,[object Object],[object Object],[object Object],[object Object],[object Object],[object Object],any_color mono_color polyo_color red   blue  green   orange   purple 
Important about hypothesis languages: ,[object Object]
Defining Concept Learning: ,[object Object],[object Object],[object Object],[object Object],[object Object],[object Object],[object Object],[object Object],[object Object],[object Object],[object Object]
The inductive learning hypothesis: If  a hypothesis approximates the target function well  over a sufficiently large number of examples , then  the hypothesis will also approximate the  target function well  on other unobserved examples .
Find-S: a naïve algorithm Replace  h  by a minimal generalization of  h  that covers  x Return   h Initialize :  h :=   For  each  positive  training example  x  in  D   Do: If   h  does  not  cover  x :
Reaction example: minimal generalizations = the individual events Generalization = replace something by ? no more positive examples: return  h Restaurant Meal Day Cost    Reaction Alma 3 breakfast Friday cheap Yes De Moete lunch Friday expensive No Alma 3 lunch Saturday cheap Yes Sedes breakfast Sunday cheap No Alma 3 breakfast Sunday expensive No + - + - - Initially: h =   Example 1: h =  [Alma 3,breakfast,Friday,cheap]  Example 2: h =  [Alma 3,  ? ,  ? ,cheap] 
Properties of Find-S ,[object Object],[object Object],Beth Jo Alex Hacker Scientist Footballplayer  H Beth can be minimally generalized in 2 ways to include a new example Jo.
Properties of Find-S (2) ,[object Object],D : Beth Alex Jo + - + Beth Jo Alex Hacker Scientist Footballplayer  H Is a wrong solution
Properties of Find-S (3) ,[object Object],[object Object],D : Beth Jo  Beth + + - D : Beth Alex  Jo + - + Jo Alex Scientist  H Beth
Nice about Find-S: ,[object Object],If  h  already covered the 20 first examples,  then  h’  will as well h        X H ,[object Object], h’   
Dual Find-S: Initialize :  h :=  [ ?, ?, .., ?] For  each  negative  training example  x  in  D   Do: If   h  does cover  x : Replace  h  by a minimal specialization of  h  that does  not  cover  x Return   h
Reaction example: Restaurant Meal Day Cost    Reaction Alma 3 breakfast Friday cheap Yes De Moete lunch Friday expensive No Alma 3 lunch Saturday cheap Yes Sedes breakfast Sunday cheap No Alma 3 breakfast Sunday expensive No + - + - - Initially: h = [ ?, ?, ?, ?] Example 1:  h= [ ?, breakfast, ?, ?] Example 2:  h= [ Alma 3, breakfast, ?, ?] Example 3:  h= [ Alma 3, breakfast, ?, cheap]
Version Spaces: the idea:  ,[object Object],[object Object],[object Object],[object Object]
Version spaces: initialization ,[object Object],G = { [ ?, ?, …, ?] } S = {    }
Negative examples: ,[object Object],Invariant : only the hypotheses  more specific  than the  ones of  G  are still possible:   they don’t cover the negative example S = {    }  G = { [ ?, ?, …, ?] } G = {h1, h2, …, hn}
Positive examples: ,[object Object],Invariant : only the hypotheses  more general  than the  ones of  S  are still possible:   they do cover the positive example G = {h1, h2, …, hn} S = {    } S = { h1’, h2’, …,hm’  }
Later: negative examples ,[object Object],Invariant : only the hypotheses  more specific  than the  ones of  G  are still possible:   they don’t cover the negative example S = { h1’, h2’, …,hm’  }  G = {h1, h2, …, hn} G = {h1, h21, h22, …, hn}
Later: positive examples ,[object Object],Invariant : only the hypotheses  more general  than the  ones of  S  are still possible:   they do cover the positive example G = {h1, h21, h22, …, hn} S = { h1’, h2’, …,hm’  }  S = { h11’, h12’, h13’, h2’, …,hm’  }
Optimization: negative: ,[object Object],…  are more general than ... G = {h1, h21, …, hn} S = { h1’, h2’, …,hm’  }  Invariant : on  S  used !
Optimization: positive: ,[object Object],…  are more specific than ... Invariant : on  G  used ! G = {h1, h21, h22, …, hn} S = { h13’, h2’, …,hm’  }
Pruning: negative examples  ,[object Object],G = {h1, h21, …, hn} S = { h1’, h3’, …,hm’  }  Invariant  only works for the previous examples, not the last one Cover the last negative example!
Pruning: positive examples  ,[object Object],G = {h1, h21, …, hn} S = { h1’, h3’, …,hm’  }  Don’t cover the last positive example
Eliminate redundant hypotheses ,[object Object],More specific than another general hypothesis Obviously also for  S  ! G = {h1, h21, …, hn} S = { h1’, h3’, …,hm’  }  Reason:   Invariant acts as a wave front : anything above  G  is not allowed. The most general elements of  G  define the real boundary
Convergence: ,[object Object],[object Object],G = {h1, h21, h22, …, hn} S = { h13’, h2’, …,hm’  }
Reaction example ,[object Object],[ ?, ?, ?, ?] Most general  Most specific
Alma3,  breakfast,  Friday,  cheap:  + ,[object Object],[ ?, ?, ?, ?]  [Alma3, breakfast, Friday, cheap]
DeMoete,  lunch,  Friday,  expensive:  - ,[object Object],[object Object],[ Alma3 , ?, ?, ?] [ DeMoete , ?, ?, ?]  [ Sedes , ?, ?, ?] [?,  breakfast , ?, ?] [?,  lunch , ?, ?] [?,  dinner , ?, ?] [?, ?,  Monday , ?] [?, ?,  Tuesday , ?] [?, ?,  Wednesday , ?] [?, ?,  Thursday , ?] [?, ?,  Friday , ?] [?, ?,  Saturday , ?] [?, ?,  Sunday , ?] [?, ?, ?,  cheap ] [?, ?, ?,  expensive ] Matches the negative example X X X X Does not generalize the specific model Specific model: [Alma3, breakfast, Friday, cheap] X X X X X X X X Remain !
Result after example 2: [ ?, ?, ?, ?]  [Alma3, breakfast, Friday, cheap] [ Alma3 , ?, ?, ?] [?,  breakfast , ?, ?] [?, ?, ?,  cheap ]
Alma3,  lunch,  Saturday,  cheap:  + ,[object Object],[Alma3, breakfast, Friday, cheap] [Alma3, ?, ?, ?] [?, breakfast, ?, ?] [?, ?, ?, cheap] [Alma3, ? , ? , cheap] does not match new example
Sedes,  breakfast,  Sunday,  cheap:  - ,[object Object],The only specialization that is introduced is  pruned, because it is more specific than  another general hypothesis [Alma3, ?, ?, ?] [?, ?, ?, cheap] [Alma3, ? , ? , cheap] [ Alma3 , ?, ?, cheap]
Alma 3,  breakfast,  Sunday,  expensive:  - ,[object Object],Cheap food at Alma3 produces the allergy ! [Alma3, ?, ?, ?] [Alma3, ? , ? , cheap] [Alma3, ?, ?,  cheap ] Same hypothesis !!!
Version Space Algorithm: Generalize  all hypotheses in  S  that do not cover the    example yet, but ensure the following:   - Only introduce  minimal changes  on the hypotheses.   -  Each  new specific hypothesis is a  specialization of   some general hypothesis .   -  No  new specific hypothesis is a  generalization of   some other specific hypothesis . Prune away  all hypotheses in  G  that do not cover the   example. Initially :  G := { the hypothesis that covers everything}   S := {  } For each new  positive  example:
Version Space Algorithm (2): Specialize  all hypotheses in  G  that cover the example,    but ensure the following:   - Only introduce  minimal changes  on the hypotheses.   -  Each  new general hypothesis is a  generalization of   some specific hypothesis .   -  No  new general hypothesis is a  specialization of   some other general hypothesis . Prune away  all hypotheses in  S  that cover the example. Until there are no more examples:   report  S  and  G OR  S  or  G  become empty :  report failure . . . For each new  negative  example:
Properties of VS: ,[object Object],[object Object],[object Object],[object Object],[object Object],[object Object]
Termination: ,[object Object],Then  all these hypotheses , and  all intermediate hypotheses , are still correct descriptions covering the test data. ,[object Object],[Alma3, ?, ?, ?] [?, ?, Monday,?] [Alma3, ?, Monday, cheap] [Alma3, ?, ?, cheap] [?, ?, Monday, cheap] [Alma3, ?, Monday, ?] VS makes NO unnecessary choices !
Termination (2): ,[object Object],[object Object],[object Object],[object Object],[Alma3, breakfast, ?, cheap]    [Alma3, lunch, ?, cheap] <Alma3, dinner, Sunday, cheap>  - <Alma3, breakfast, Sunday, cheap>  + <Alma3, lunch, Sunday, cheap>  + ,[object Object],[object Object],[object Object],[object Object]
Which example next? ,[object Object],[object Object],[object Object],<Alma3,lunch,Monday, expensive >   [Alma3, ?, ?, ?] [?, ?, Monday,?] [Alma3, ?, Monday, cheap] [Alma3, ?, ?, cheap] [?, ?, Monday, cheap] [Alma3, ?, Monday, ?] classified negative by 3 hypotheses classified positive by 3 hypotheses is the most informative new example
Use of partially learned concepts: ,[object Object],[Alma3, ?, ?, ?] [?, ?, Monday,?] [Alma3, ?, Monday, cheap] [Alma3, ?, ?, cheap] [?, ?, Monday, cheap] [Alma3, ?, Monday, ?] ,[object Object],[object Object],[object Object]
Use of partially learned concepts (2): ,[object Object],[Alma3, ?, ?, ?] [?, ?, Monday,?] [Alma3, ?, Monday, cheap] [Alma3, ?, ?, cheap] [?, ?, Monday, cheap] [Alma3, ?, Monday, ?] ,[object Object],[object Object],[object Object]
Use of partially learned concepts (3): ,[object Object],[Alma3, ?, ?, ?] [?, ?, Monday,?] [Alma3, ?, Monday, cheap] [Alma3, ?, ?, cheap] [?, ?, Monday, cheap] [Alma3, ?, Monday, ?] ,[object Object],[object Object],[object Object]
Use of partially learned concepts (4): ,[object Object],[Alma3, ?, ?, ?] [?, ?, Monday,?] [Alma3, ?, Monday, cheap] [Alma3, ?, ?, cheap] [?, ?, Monday, cheap] [Alma3, ?, Monday, ?] probably does not belong to the concept : ratio : 1/6 ,[object Object],[object Object]
The relevance of inductive BIAS: choosing H ,[object Object],[object Object],[object Object],[object Object],[object Object],[object Object],[object Object],[object Object]
Inductive BIAS (2): ,[object Object],[object Object],[object Object],Restaurant Meal Day Cost    3  X  3  X  7  X  2  =  126
Inductive BIAS (3) ,[object Object],Restaurant Meal Day Cost    Reaction Alma 3 breakfast Friday cheap Yes De Moete lunch Friday expensive No Alma 3 lunch Saturday cheap Yes Sedes breakfast Sunday cheap No Alma 3 breakfast Sunday expensive No + - + - - ?  {~[DeMoete. Lunch, Friday, expensive]    ~[Sedes, breakfast, Sunday, cheap]} {~[DeMoete. Lunch, Friday, expensive]    ~[Sedes, breakfast, Sunday, cheap]      ~[Alma 3, breakfast, Sunday, expensive]} {~[DeMoete. Lunch, Friday, expensive]} {[Alma 3, breakfast, Friday, cheap]   [Alma 3, lunch, Saturday, cheap]} {[Alma 3, breakfast, Friday, cheap]}
Inductive BIAS (3) ,[object Object],[object Object],[object Object],G = {~[DeMoete. Lunch, Friday, expensive]     ~[Sedes, breakfast, Sunday, cheap]    ~[Alma 3, breakfast, Sunday, expensive]} S = {[Alma 3, breakfast, Friday, cheap]    [Alma 3, lunch, Saturday, cheap]} ,[object Object],[object Object]
Shift of Bias ,[object Object],[object Object],[object Object],[object Object],[object Object],[object Object],[object Object]

More Related Content

Viewers also liked

ICF - EE in HOME Workshop
ICF - EE in HOME WorkshopICF - EE in HOME Workshop
ICF - EE in HOME WorkshopICF_HCD
 
Unit 1.1 Introduction to Programming
Unit 1.1 Introduction to Programming Unit 1.1 Introduction to Programming
Unit 1.1 Introduction to Programming Intan Jameel
 
Addict Wipii English 081128
Addict Wipii English 081128Addict Wipii English 081128
Addict Wipii English 081128eardigo
 
D2.3.1 Evaluation results of the LinkedUp Veni competition
D2.3.1 Evaluation results of the LinkedUp Veni competitionD2.3.1 Evaluation results of the LinkedUp Veni competition
D2.3.1 Evaluation results of the LinkedUp Veni competitionHendrik Drachsler
 
Cisco routingtable2 lookup
Cisco routingtable2 lookupCisco routingtable2 lookup
Cisco routingtable2 lookupMakram Ghazzaoui
 
Product Development at the Smithsonian Libraries
Product Development at the Smithsonian LibrariesProduct Development at the Smithsonian Libraries
Product Development at the Smithsonian Librarieseclemrush
 
A Long Walk to Water - Lssn 14
A Long Walk to Water - Lssn 14A Long Walk to Water - Lssn 14
A Long Walk to Water - Lssn 14Terri Weiss
 
Winners of HTML5 BYOG - InGDIn
Winners of HTML5 BYOG - InGDInWinners of HTML5 BYOG - InGDIn
Winners of HTML5 BYOG - InGDInKinshuk Sunil
 
Inbound Marketing Explained
Inbound Marketing ExplainedInbound Marketing Explained
Inbound Marketing ExplainedEddie Choi
 
Language learning and teaching through social media
Language learning and teaching through social mediaLanguage learning and teaching through social media
Language learning and teaching through social mediaLis Parcell
 
Adriano Augusto Michieli, Enrico Stanley, Genova, Formiggini, 1913
Adriano Augusto Michieli, Enrico Stanley, Genova, Formiggini, 1913Adriano Augusto Michieli, Enrico Stanley, Genova, Formiggini, 1913
Adriano Augusto Michieli, Enrico Stanley, Genova, Formiggini, 1913stefanogambari
 
JISC Greening ICT Keynote Presentation
JISC Greening ICT Keynote PresentationJISC Greening ICT Keynote Presentation
JISC Greening ICT Keynote PresentationJoss Winn
 
Resourcing the digital learner
Resourcing the digital learnerResourcing the digital learner
Resourcing the digital learnerLis Parcell
 
The changing landscape, a personal view for OER Wales
The changing landscape, a personal view for OER WalesThe changing landscape, a personal view for OER Wales
The changing landscape, a personal view for OER WalesLis Parcell
 
12 13 Initiative: Breakout Session OHIONET Annual Meeting 2009
12 13 Initiative: Breakout Session OHIONET Annual Meeting 200912 13 Initiative: Breakout Session OHIONET Annual Meeting 2009
12 13 Initiative: Breakout Session OHIONET Annual Meeting 2009ohionet
 

Viewers also liked (20)

ICF - EE in HOME Workshop
ICF - EE in HOME WorkshopICF - EE in HOME Workshop
ICF - EE in HOME Workshop
 
Unit 2.5
Unit 2.5Unit 2.5
Unit 2.5
 
Unit 1.1 Introduction to Programming
Unit 1.1 Introduction to Programming Unit 1.1 Introduction to Programming
Unit 1.1 Introduction to Programming
 
Addict Wipii English 081128
Addict Wipii English 081128Addict Wipii English 081128
Addict Wipii English 081128
 
D2.3.1 Evaluation results of the LinkedUp Veni competition
D2.3.1 Evaluation results of the LinkedUp Veni competitionD2.3.1 Evaluation results of the LinkedUp Veni competition
D2.3.1 Evaluation results of the LinkedUp Veni competition
 
Mwomen
MwomenMwomen
Mwomen
 
Cisco routingtable2 lookup
Cisco routingtable2 lookupCisco routingtable2 lookup
Cisco routingtable2 lookup
 
Product Development at the Smithsonian Libraries
Product Development at the Smithsonian LibrariesProduct Development at the Smithsonian Libraries
Product Development at the Smithsonian Libraries
 
A Long Walk to Water - Lssn 14
A Long Walk to Water - Lssn 14A Long Walk to Water - Lssn 14
A Long Walk to Water - Lssn 14
 
Winners of HTML5 BYOG - InGDIn
Winners of HTML5 BYOG - InGDInWinners of HTML5 BYOG - InGDIn
Winners of HTML5 BYOG - InGDIn
 
Inbound Marketing Explained
Inbound Marketing ExplainedInbound Marketing Explained
Inbound Marketing Explained
 
Unit 2.12
Unit 2.12Unit 2.12
Unit 2.12
 
Language learning and teaching through social media
Language learning and teaching through social mediaLanguage learning and teaching through social media
Language learning and teaching through social media
 
Corel
CorelCorel
Corel
 
Adriano Augusto Michieli, Enrico Stanley, Genova, Formiggini, 1913
Adriano Augusto Michieli, Enrico Stanley, Genova, Formiggini, 1913Adriano Augusto Michieli, Enrico Stanley, Genova, Formiggini, 1913
Adriano Augusto Michieli, Enrico Stanley, Genova, Formiggini, 1913
 
M02 un09 p01
M02 un09 p01M02 un09 p01
M02 un09 p01
 
JISC Greening ICT Keynote Presentation
JISC Greening ICT Keynote PresentationJISC Greening ICT Keynote Presentation
JISC Greening ICT Keynote Presentation
 
Resourcing the digital learner
Resourcing the digital learnerResourcing the digital learner
Resourcing the digital learner
 
The changing landscape, a personal view for OER Wales
The changing landscape, a personal view for OER WalesThe changing landscape, a personal view for OER Wales
The changing landscape, a personal view for OER Wales
 
12 13 Initiative: Breakout Session OHIONET Annual Meeting 2009
12 13 Initiative: Breakout Session OHIONET Annual Meeting 200912 13 Initiative: Breakout Session OHIONET Annual Meeting 2009
12 13 Initiative: Breakout Session OHIONET Annual Meeting 2009
 

Similar to Version spaces

concept-learning.ppt
concept-learning.pptconcept-learning.ppt
concept-learning.pptpatel252389
 
concept-learning of artificial intelligence
concept-learning of artificial intelligenceconcept-learning of artificial intelligence
concept-learning of artificial intelligencessuser01fa1b
 
ML_Unit_1_Part_B
ML_Unit_1_Part_BML_Unit_1_Part_B
ML_Unit_1_Part_BSrimatre K
 
Concept learning and candidate elimination algorithm
Concept learning and candidate elimination algorithmConcept learning and candidate elimination algorithm
Concept learning and candidate elimination algorithmswapnac12
 
Lecture 7
Lecture 7Lecture 7
Lecture 7butest
 
Lecture 7
Lecture 7Lecture 7
Lecture 7butest
 
artificial intelligence.pptx
artificial intelligence.pptxartificial intelligence.pptx
artificial intelligence.pptxSabthamiS1
 
Cs229 notes4
Cs229 notes4Cs229 notes4
Cs229 notes4VuTran231
 
Bayesian Learning- part of machine learning
Bayesian Learning-  part of machine learningBayesian Learning-  part of machine learning
Bayesian Learning- part of machine learningkensaleste
 

Similar to Version spaces (11)

concept-learning.ppt
concept-learning.pptconcept-learning.ppt
concept-learning.ppt
 
concept-learning of artificial intelligence
concept-learning of artificial intelligenceconcept-learning of artificial intelligence
concept-learning of artificial intelligence
 
ML_Unit_1_Part_B
ML_Unit_1_Part_BML_Unit_1_Part_B
ML_Unit_1_Part_B
 
Concept learning and candidate elimination algorithm
Concept learning and candidate elimination algorithmConcept learning and candidate elimination algorithm
Concept learning and candidate elimination algorithm
 
Lecture 7
Lecture 7Lecture 7
Lecture 7
 
Lecture 7
Lecture 7Lecture 7
Lecture 7
 
Lec 17
Lec  17Lec  17
Lec 17
 
artificial intelligence.pptx
artificial intelligence.pptxartificial intelligence.pptx
artificial intelligence.pptx
 
Cs229 notes4
Cs229 notes4Cs229 notes4
Cs229 notes4
 
Bayesian Learning- part of machine learning
Bayesian Learning-  part of machine learningBayesian Learning-  part of machine learning
Bayesian Learning- part of machine learning
 
ML02.ppt
ML02.pptML02.ppt
ML02.ppt
 

Recently uploaded

What's New in Teams Calling, Meetings and Devices March 2024
What's New in Teams Calling, Meetings and Devices March 2024What's New in Teams Calling, Meetings and Devices March 2024
What's New in Teams Calling, Meetings and Devices March 2024Stephanie Beckett
 
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptxUse of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptxLoriGlavin3
 
Gen AI in Business - Global Trends Report 2024.pdf
Gen AI in Business - Global Trends Report 2024.pdfGen AI in Business - Global Trends Report 2024.pdf
Gen AI in Business - Global Trends Report 2024.pdfAddepto
 
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptxPasskey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptxLoriGlavin3
 
DSPy a system for AI to Write Prompts and Do Fine Tuning
DSPy a system for AI to Write Prompts and Do Fine TuningDSPy a system for AI to Write Prompts and Do Fine Tuning
DSPy a system for AI to Write Prompts and Do Fine TuningLars Bell
 
TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024Lonnie McRorey
 
Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)
Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)
Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)Mark Simos
 
Generative AI for Technical Writer or Information Developers
Generative AI for Technical Writer or Information DevelopersGenerative AI for Technical Writer or Information Developers
Generative AI for Technical Writer or Information DevelopersRaghuram Pandurangan
 
"Debugging python applications inside k8s environment", Andrii Soldatenko
"Debugging python applications inside k8s environment", Andrii Soldatenko"Debugging python applications inside k8s environment", Andrii Soldatenko
"Debugging python applications inside k8s environment", Andrii SoldatenkoFwdays
 
From Family Reminiscence to Scholarly Archive .
From Family Reminiscence to Scholarly Archive .From Family Reminiscence to Scholarly Archive .
From Family Reminiscence to Scholarly Archive .Alan Dix
 
TrustArc Webinar - How to Build Consumer Trust Through Data Privacy
TrustArc Webinar - How to Build Consumer Trust Through Data PrivacyTrustArc Webinar - How to Build Consumer Trust Through Data Privacy
TrustArc Webinar - How to Build Consumer Trust Through Data PrivacyTrustArc
 
Connect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck PresentationConnect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck PresentationSlibray Presentation
 
Unleash Your Potential - Namagunga Girls Coding Club
Unleash Your Potential - Namagunga Girls Coding ClubUnleash Your Potential - Namagunga Girls Coding Club
Unleash Your Potential - Namagunga Girls Coding ClubKalema Edgar
 
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek SchlawackFwdays
 
Dev Dives: Streamline document processing with UiPath Studio Web
Dev Dives: Streamline document processing with UiPath Studio WebDev Dives: Streamline document processing with UiPath Studio Web
Dev Dives: Streamline document processing with UiPath Studio WebUiPathCommunity
 
unit 4 immunoblotting technique complete.pptx
unit 4 immunoblotting technique complete.pptxunit 4 immunoblotting technique complete.pptx
unit 4 immunoblotting technique complete.pptxBkGupta21
 
WordPress Websites for Engineers: Elevate Your Brand
WordPress Websites for Engineers: Elevate Your BrandWordPress Websites for Engineers: Elevate Your Brand
WordPress Websites for Engineers: Elevate Your Brandgvaughan
 
SALESFORCE EDUCATION CLOUD | FEXLE SERVICES
SALESFORCE EDUCATION CLOUD | FEXLE SERVICESSALESFORCE EDUCATION CLOUD | FEXLE SERVICES
SALESFORCE EDUCATION CLOUD | FEXLE SERVICESmohitsingh558521
 
What is DBT - The Ultimate Data Build Tool.pdf
What is DBT - The Ultimate Data Build Tool.pdfWhat is DBT - The Ultimate Data Build Tool.pdf
What is DBT - The Ultimate Data Build Tool.pdfMounikaPolabathina
 

Recently uploaded (20)

What's New in Teams Calling, Meetings and Devices March 2024
What's New in Teams Calling, Meetings and Devices March 2024What's New in Teams Calling, Meetings and Devices March 2024
What's New in Teams Calling, Meetings and Devices March 2024
 
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptxUse of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
 
Gen AI in Business - Global Trends Report 2024.pdf
Gen AI in Business - Global Trends Report 2024.pdfGen AI in Business - Global Trends Report 2024.pdf
Gen AI in Business - Global Trends Report 2024.pdf
 
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptxPasskey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptx
 
DMCC Future of Trade Web3 - Special Edition
DMCC Future of Trade Web3 - Special EditionDMCC Future of Trade Web3 - Special Edition
DMCC Future of Trade Web3 - Special Edition
 
DSPy a system for AI to Write Prompts and Do Fine Tuning
DSPy a system for AI to Write Prompts and Do Fine TuningDSPy a system for AI to Write Prompts and Do Fine Tuning
DSPy a system for AI to Write Prompts and Do Fine Tuning
 
TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024
 
Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)
Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)
Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)
 
Generative AI for Technical Writer or Information Developers
Generative AI for Technical Writer or Information DevelopersGenerative AI for Technical Writer or Information Developers
Generative AI for Technical Writer or Information Developers
 
"Debugging python applications inside k8s environment", Andrii Soldatenko
"Debugging python applications inside k8s environment", Andrii Soldatenko"Debugging python applications inside k8s environment", Andrii Soldatenko
"Debugging python applications inside k8s environment", Andrii Soldatenko
 
From Family Reminiscence to Scholarly Archive .
From Family Reminiscence to Scholarly Archive .From Family Reminiscence to Scholarly Archive .
From Family Reminiscence to Scholarly Archive .
 
TrustArc Webinar - How to Build Consumer Trust Through Data Privacy
TrustArc Webinar - How to Build Consumer Trust Through Data PrivacyTrustArc Webinar - How to Build Consumer Trust Through Data Privacy
TrustArc Webinar - How to Build Consumer Trust Through Data Privacy
 
Connect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck PresentationConnect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck Presentation
 
Unleash Your Potential - Namagunga Girls Coding Club
Unleash Your Potential - Namagunga Girls Coding ClubUnleash Your Potential - Namagunga Girls Coding Club
Unleash Your Potential - Namagunga Girls Coding Club
 
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
 
Dev Dives: Streamline document processing with UiPath Studio Web
Dev Dives: Streamline document processing with UiPath Studio WebDev Dives: Streamline document processing with UiPath Studio Web
Dev Dives: Streamline document processing with UiPath Studio Web
 
unit 4 immunoblotting technique complete.pptx
unit 4 immunoblotting technique complete.pptxunit 4 immunoblotting technique complete.pptx
unit 4 immunoblotting technique complete.pptx
 
WordPress Websites for Engineers: Elevate Your Brand
WordPress Websites for Engineers: Elevate Your BrandWordPress Websites for Engineers: Elevate Your Brand
WordPress Websites for Engineers: Elevate Your Brand
 
SALESFORCE EDUCATION CLOUD | FEXLE SERVICES
SALESFORCE EDUCATION CLOUD | FEXLE SERVICESSALESFORCE EDUCATION CLOUD | FEXLE SERVICES
SALESFORCE EDUCATION CLOUD | FEXLE SERVICES
 
What is DBT - The Ultimate Data Build Tool.pdf
What is DBT - The Ultimate Data Build Tool.pdfWhat is DBT - The Ultimate Data Build Tool.pdf
What is DBT - The Ultimate Data Build Tool.pdf
 

Version spaces

  • 1. Machine Learning Version Spaces Learning
  • 2.
  • 3. Version Spaces A concept learning technique based on refining models of the world.
  • 4.
  • 5.
  • 6. Pictured: Set of all possible events                                   Given some examples: + + + + - - - - Find a concept that covers all the positive examples and none of the negative ones!
  • 7.
  • 8.
  • 9.
  • 10.
  • 11.
  • 12.
  • 13.
  • 14. Hypotheses relate to sets of possible events       Events    x2      x1 x1 = < Alma 3, lunch, Monday, expensive> x2 = < Sedes, lunch, Sunday, cheap> h1 = [?, lunch, Monday, ?] h2 = [?, lunch, ?, cheap] h3 = [?, lunch, ?, ?] General Specific                 Hypotheses h2 h3 h1
  • 15.
  • 16.
  • 17.
  • 18.
  • 19. The inductive learning hypothesis: If a hypothesis approximates the target function well over a sufficiently large number of examples , then the hypothesis will also approximate the target function well on other unobserved examples .
  • 20. Find-S: a naïve algorithm Replace h by a minimal generalization of h that covers x Return h Initialize : h :=  For each positive training example x in D Do: If h does not cover x :
  • 21. Reaction example: minimal generalizations = the individual events Generalization = replace something by ? no more positive examples: return h Restaurant Meal Day Cost Reaction Alma 3 breakfast Friday cheap Yes De Moete lunch Friday expensive No Alma 3 lunch Saturday cheap Yes Sedes breakfast Sunday cheap No Alma 3 breakfast Sunday expensive No + - + - - Initially: h =  Example 1: h = [Alma 3,breakfast,Friday,cheap]  Example 2: h = [Alma 3, ? , ? ,cheap] 
  • 22.
  • 23.
  • 24.
  • 25.
  • 26. Dual Find-S: Initialize : h := [ ?, ?, .., ?] For each negative training example x in D Do: If h does cover x : Replace h by a minimal specialization of h that does not cover x Return h
  • 27. Reaction example: Restaurant Meal Day Cost Reaction Alma 3 breakfast Friday cheap Yes De Moete lunch Friday expensive No Alma 3 lunch Saturday cheap Yes Sedes breakfast Sunday cheap No Alma 3 breakfast Sunday expensive No + - + - - Initially: h = [ ?, ?, ?, ?] Example 1:  h= [ ?, breakfast, ?, ?] Example 2:  h= [ Alma 3, breakfast, ?, ?] Example 3:  h= [ Alma 3, breakfast, ?, cheap]
  • 28.
  • 29.
  • 30.
  • 31.
  • 32.
  • 33.
  • 34.
  • 35.
  • 36.
  • 37.
  • 38.
  • 39.
  • 40.
  • 41.
  • 42.
  • 43. Result after example 2: [ ?, ?, ?, ?]  [Alma3, breakfast, Friday, cheap] [ Alma3 , ?, ?, ?] [?, breakfast , ?, ?] [?, ?, ?, cheap ]
  • 44.
  • 45.
  • 46.
  • 47. Version Space Algorithm: Generalize all hypotheses in S that do not cover the example yet, but ensure the following: - Only introduce minimal changes on the hypotheses. - Each new specific hypothesis is a specialization of some general hypothesis . - No new specific hypothesis is a generalization of some other specific hypothesis . Prune away all hypotheses in G that do not cover the example. Initially : G := { the hypothesis that covers everything} S := {  } For each new positive example:
  • 48. Version Space Algorithm (2): Specialize all hypotheses in G that cover the example, but ensure the following: - Only introduce minimal changes on the hypotheses. - Each new general hypothesis is a generalization of some specific hypothesis . - No new general hypothesis is a specialization of some other general hypothesis . Prune away all hypotheses in S that cover the example. Until there are no more examples: report S and G OR S or G become empty : report failure . . . For each new negative example:
  • 49.
  • 50.
  • 51.
  • 52.
  • 53.
  • 54.
  • 55.
  • 56.
  • 57.
  • 58.
  • 59.
  • 60.
  • 61.