SlideShare une entreprise Scribd logo
1  sur  49
Lukas Biewald
The Effect of Better Algorithms
0%
5%
10%
15%
20%
25%
Naïve Bayes Maximum
Entropy
SVM
Classifier Error Rate
Active Semi-Supervised Learning for Improving
Word Alignment
(Vamshi ACL ’10)
Real World Data
2
The Effect of Better Features
0%
5%
10%
15%
20%
25%
30%
Unigrams Bigrams Unigrams+Bigrams
Classifier Error Rate
3
The Effect of More Data
Active Semi-Supervised Learning for
Improving Word Alignment
(Vamshi ACL ’10)
Real World Data
0%
2%
4%
6%
8%
10%
12%
14%
N 2N 4N
Classifier Error Rate
4
The Effect of Cleaner Data
0%
2%
4%
6%
8%
10%
12%
14%
90% Accurate Data 95% Accurate Data 100% Accurate Data
Classifier Error Rate
5
Where Do Data Scientists Spend Their Time?
6
Source: CrowdFlower Data
Science Report 2015
CrowdFlower Data Enrichment Platform
7
Color Data
8
9
10
11
12
13
14
Apple Watch
15
Apple Watch
16
Apple Watch
17
Apple Watch
18
Collecting the Same Data Over and Over
19
Open Data
20
Make Your Data Public Setting
21
Data for Everyone
22
Data For Everyone Library
23
Data for Everyone
24
Data For Everyone
25
Open Data API
26
URL Categorization
27
Categorize URLs
28
Record Data
29
Extracting Names and Titles
30
Summarization
31
Is an Image Funny?
32
Classifying Medical Images
33
Attributes of People
34
35
Kaggle accuracy
0%
10%
20%
30%
40%
50%
60%
70%
Baseline 12-May 13-May 14-May 15-May
Accuracy
Accuracy of Best Performing Model
36
Kaggle accuracy over time
0%
10%
20%
30%
40%
50%
60%
70%
80%
Accuracy
Accuracy of the Best Performing Model
37
Kaggle Participation
0
200
400
600
800
1000
1200
1400
Number of Participating Teams
38
AI
Classifier
Output
Confident
Human in the Loop
39
Human in the Loop
Confident
Output
AI
Classifier
Human
Annotation
40
Human in the Loop
Confident
Output
AI
Classifier
Active Learning
Human
Annotation
41
Human in the Loop
Confident
Output
AI
Classifier
Active Learning
Human
Annotation
42
Active Learning
From hunch.net active learning tutorial ICML ‘09
43
Active Learning Accuracy Improvement
44
Google Cars Miles Per Disengage
45
Adaptive Cruise Control
Image source: ExtremeTech
46
Advanced Chess
Image source: Computer Chess
47
AlphaGo
48
Lukas Biewald
lukas@crowdflower.com
@L2K
Thank You

Contenu connexe

Similaire à Active Learning and Human-in-the-Loop

UX by the numbers: Discovering the why from numbers
UX by the numbers: Discovering the why from numbersUX by the numbers: Discovering the why from numbers
UX by the numbers: Discovering the why from numbers
UXPA UK
 
[TestWarez 2017] Od testowania do monitoringu jakości – wyzwania Continuous ...
[TestWarez 2017]  Od testowania do monitoringu jakości – wyzwania Continuous ...[TestWarez 2017]  Od testowania do monitoringu jakości – wyzwania Continuous ...
[TestWarez 2017] Od testowania do monitoringu jakości – wyzwania Continuous ...
Stowarzyszenie Jakości Systemów Informatycznych (SJSI)
 

Similaire à Active Learning and Human-in-the-Loop (20)

Cloud VM World Tradeshow Report
Cloud VM World Tradeshow ReportCloud VM World Tradeshow Report
Cloud VM World Tradeshow Report
 
How to Enter the Data Analytics Industry?
How to Enter the Data Analytics Industry?How to Enter the Data Analytics Industry?
How to Enter the Data Analytics Industry?
 
Managing Unrealistic User Expectations (i.e Digital for Spoilt Brats)
Managing Unrealistic User Expectations (i.e Digital for Spoilt Brats)Managing Unrealistic User Expectations (i.e Digital for Spoilt Brats)
Managing Unrealistic User Expectations (i.e Digital for Spoilt Brats)
 
Digital Disruption Asia - Pleasing the Unpleasable with Digital Performance Data
Digital Disruption Asia - Pleasing the Unpleasable with Digital Performance DataDigital Disruption Asia - Pleasing the Unpleasable with Digital Performance Data
Digital Disruption Asia - Pleasing the Unpleasable with Digital Performance Data
 
The 2016 Watch List
The 2016 Watch ListThe 2016 Watch List
The 2016 Watch List
 
How To Plan for, Design, and Implement a Communications Dashboard
How To Plan for, Design, and Implement a Communications DashboardHow To Plan for, Design, and Implement a Communications Dashboard
How To Plan for, Design, and Implement a Communications Dashboard
 
2020 Testing Trends: Top Predictions for QA Teams to Watch, Join, and Lead
2020 Testing Trends: Top Predictions for QA Teams to Watch, Join, and Lead2020 Testing Trends: Top Predictions for QA Teams to Watch, Join, and Lead
2020 Testing Trends: Top Predictions for QA Teams to Watch, Join, and Lead
 
Early Lessons Learned in Applying Big Data To TV Advertising
Early Lessons Learned in Applying Big Data To TV AdvertisingEarly Lessons Learned in Applying Big Data To TV Advertising
Early Lessons Learned in Applying Big Data To TV Advertising
 
Early Lessons Learned in Applying Big Data To TV Advertising
Early Lessons Learned in Applying Big Data To TV AdvertisingEarly Lessons Learned in Applying Big Data To TV Advertising
Early Lessons Learned in Applying Big Data To TV Advertising
 
Heavy, Messy, Misleading: How Big Data is a human problem, not a tech one
Heavy, Messy, Misleading: How Big Data is a human problem, not a tech oneHeavy, Messy, Misleading: How Big Data is a human problem, not a tech one
Heavy, Messy, Misleading: How Big Data is a human problem, not a tech one
 
Population Stability Index(PSI) for Big Data World
Population Stability Index(PSI) for Big Data WorldPopulation Stability Index(PSI) for Big Data World
Population Stability Index(PSI) for Big Data World
 
Digital disruption
Digital disruptionDigital disruption
Digital disruption
 
JDO 2019: Data Science for Developers - Matthew Renze
JDO 2019: Data Science for Developers -  Matthew RenzeJDO 2019: Data Science for Developers -  Matthew Renze
JDO 2019: Data Science for Developers - Matthew Renze
 
FIAT/IFTA MAM Survey - Results overview
FIAT/IFTA MAM Survey - Results overviewFIAT/IFTA MAM Survey - Results overview
FIAT/IFTA MAM Survey - Results overview
 
UX by the numbers: Discovering the why from numbers
UX by the numbers: Discovering the why from numbersUX by the numbers: Discovering the why from numbers
UX by the numbers: Discovering the why from numbers
 
Discovering WHY from numbers
Discovering WHY from numbersDiscovering WHY from numbers
Discovering WHY from numbers
 
Measuring Video Effectiveness - Wistiafest 2015
Measuring Video Effectiveness - Wistiafest 2015Measuring Video Effectiveness - Wistiafest 2015
Measuring Video Effectiveness - Wistiafest 2015
 
[TestWarez 2017] Od testowania do monitoringu jakości – wyzwania Continuous ...
[TestWarez 2017]  Od testowania do monitoringu jakości – wyzwania Continuous ...[TestWarez 2017]  Od testowania do monitoringu jakości – wyzwania Continuous ...
[TestWarez 2017] Od testowania do monitoringu jakości – wyzwania Continuous ...
 
Pitch Deck Teardown: Lumigo's $29 million Series A deck
Pitch Deck Teardown: Lumigo's $29 million Series A deckPitch Deck Teardown: Lumigo's $29 million Series A deck
Pitch Deck Teardown: Lumigo's $29 million Series A deck
 
ITESOFT Insider View | The Importance of Capture
ITESOFT Insider View | The Importance of CaptureITESOFT Insider View | The Importance of Capture
ITESOFT Insider View | The Importance of Capture
 

Plus de CrowdFlower

Expert Crowdsourcing with Flash Teams | CrowdConf 2013 poster
Expert Crowdsourcing with Flash Teams | CrowdConf 2013 posterExpert Crowdsourcing with Flash Teams | CrowdConf 2013 poster
Expert Crowdsourcing with Flash Teams | CrowdConf 2013 poster
CrowdFlower
 
The State of Enterprise Crowdsourcing 2013
The State of Enterprise Crowdsourcing 2013The State of Enterprise Crowdsourcing 2013
The State of Enterprise Crowdsourcing 2013
CrowdFlower
 

Plus de CrowdFlower (11)

7 Myths of AI
7 Myths of AI7 Myths of AI
7 Myths of AI
 
Building Better Models Faster Using Active Learning
Building Better Models Faster Using Active LearningBuilding Better Models Faster Using Active Learning
Building Better Models Faster Using Active Learning
 
CrowdFlower NDA Crowds - Secure, exceptional tasking at a massive scale.
CrowdFlower NDA Crowds - Secure, exceptional tasking at a massive scale. CrowdFlower NDA Crowds - Secure, exceptional tasking at a massive scale.
CrowdFlower NDA Crowds - Secure, exceptional tasking at a massive scale.
 
CrowdFlower Product Webinar - Graphical Editor and Visual Reports
CrowdFlower Product Webinar - Graphical Editor and Visual ReportsCrowdFlower Product Webinar - Graphical Editor and Visual Reports
CrowdFlower Product Webinar - Graphical Editor and Visual Reports
 
How Oracle Uses CrowdFlower For Sentiment Analysis
How Oracle Uses CrowdFlower For Sentiment AnalysisHow Oracle Uses CrowdFlower For Sentiment Analysis
How Oracle Uses CrowdFlower For Sentiment Analysis
 
Humanizing The Machine
Humanizing The MachineHumanizing The Machine
Humanizing The Machine
 
Productive Out-of-the-Box | Tooling with Yeoman to Rapidly Develop Ember.js A...
Productive Out-of-the-Box | Tooling with Yeoman to Rapidly Develop Ember.js A...Productive Out-of-the-Box | Tooling with Yeoman to Rapidly Develop Ember.js A...
Productive Out-of-the-Box | Tooling with Yeoman to Rapidly Develop Ember.js A...
 
Virtual Data Steward: Data Management 3.0
Virtual Data Steward: Data Management 3.0Virtual Data Steward: Data Management 3.0
Virtual Data Steward: Data Management 3.0
 
Expert Crowdsourcing with Flash Teams | CrowdConf 2013 poster
Expert Crowdsourcing with Flash Teams | CrowdConf 2013 posterExpert Crowdsourcing with Flash Teams | CrowdConf 2013 poster
Expert Crowdsourcing with Flash Teams | CrowdConf 2013 poster
 
The State of Enterprise Crowdsourcing 2013
The State of Enterprise Crowdsourcing 2013The State of Enterprise Crowdsourcing 2013
The State of Enterprise Crowdsourcing 2013
 
CrowdFlower University Oct. 21 2013
CrowdFlower University Oct. 21 2013CrowdFlower University Oct. 21 2013
CrowdFlower University Oct. 21 2013
 

Dernier

Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
amitlee9823
 
Schema on read is obsolete. Welcome metaprogramming..pdf
Schema on read is obsolete. Welcome metaprogramming..pdfSchema on read is obsolete. Welcome metaprogramming..pdf
Schema on read is obsolete. Welcome metaprogramming..pdf
Lars Albertsson
 
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
amitlee9823
 
Delhi 99530 vip 56974 Genuine Escort Service Call Girls in Kishangarh
Delhi 99530 vip 56974 Genuine Escort Service Call Girls in  KishangarhDelhi 99530 vip 56974 Genuine Escort Service Call Girls in  Kishangarh
Delhi 99530 vip 56974 Genuine Escort Service Call Girls in Kishangarh
9953056974 Low Rate Call Girls In Saket, Delhi NCR
 
Call Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts Service
Call Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts ServiceCall Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts Service
Call Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts Service
9953056974 Low Rate Call Girls In Saket, Delhi NCR
 
Delhi Call Girls CP 9711199171 ☎✔👌✔ Whatsapp Hard And Sexy Vip Call
Delhi Call Girls CP 9711199171 ☎✔👌✔ Whatsapp Hard And Sexy Vip CallDelhi Call Girls CP 9711199171 ☎✔👌✔ Whatsapp Hard And Sexy Vip Call
Delhi Call Girls CP 9711199171 ☎✔👌✔ Whatsapp Hard And Sexy Vip Call
shivangimorya083
 
Determinants of health, dimensions of health, positive health and spectrum of...
Determinants of health, dimensions of health, positive health and spectrum of...Determinants of health, dimensions of health, positive health and spectrum of...
Determinants of health, dimensions of health, positive health and spectrum of...
shambhavirathore45
 
Al Barsha Escorts $#$ O565212860 $#$ Escort Service In Al Barsha
Al Barsha Escorts $#$ O565212860 $#$ Escort Service In Al BarshaAl Barsha Escorts $#$ O565212860 $#$ Escort Service In Al Barsha
Al Barsha Escorts $#$ O565212860 $#$ Escort Service In Al Barsha
AroojKhan71
 
Vip Model Call Girls (Delhi) Karol Bagh 9711199171✔️Body to body massage wit...
Vip Model  Call Girls (Delhi) Karol Bagh 9711199171✔️Body to body massage wit...Vip Model  Call Girls (Delhi) Karol Bagh 9711199171✔️Body to body massage wit...
Vip Model Call Girls (Delhi) Karol Bagh 9711199171✔️Body to body massage wit...
shivangimorya083
 

Dernier (20)

Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
 
Smarteg dropshipping via API with DroFx.pptx
Smarteg dropshipping via API with DroFx.pptxSmarteg dropshipping via API with DroFx.pptx
Smarteg dropshipping via API with DroFx.pptx
 
Schema on read is obsolete. Welcome metaprogramming..pdf
Schema on read is obsolete. Welcome metaprogramming..pdfSchema on read is obsolete. Welcome metaprogramming..pdf
Schema on read is obsolete. Welcome metaprogramming..pdf
 
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
 
Data-Analysis for Chicago Crime Data 2023
Data-Analysis for Chicago Crime Data  2023Data-Analysis for Chicago Crime Data  2023
Data-Analysis for Chicago Crime Data 2023
 
Delhi 99530 vip 56974 Genuine Escort Service Call Girls in Kishangarh
Delhi 99530 vip 56974 Genuine Escort Service Call Girls in  KishangarhDelhi 99530 vip 56974 Genuine Escort Service Call Girls in  Kishangarh
Delhi 99530 vip 56974 Genuine Escort Service Call Girls in Kishangarh
 
Call Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts Service
Call Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts ServiceCall Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts Service
Call Girls In Shalimar Bagh ( Delhi) 9953330565 Escorts Service
 
Delhi Call Girls CP 9711199171 ☎✔👌✔ Whatsapp Hard And Sexy Vip Call
Delhi Call Girls CP 9711199171 ☎✔👌✔ Whatsapp Hard And Sexy Vip CallDelhi Call Girls CP 9711199171 ☎✔👌✔ Whatsapp Hard And Sexy Vip Call
Delhi Call Girls CP 9711199171 ☎✔👌✔ Whatsapp Hard And Sexy Vip Call
 
Determinants of health, dimensions of health, positive health and spectrum of...
Determinants of health, dimensions of health, positive health and spectrum of...Determinants of health, dimensions of health, positive health and spectrum of...
Determinants of health, dimensions of health, positive health and spectrum of...
 
Invezz.com - Grow your wealth with trading signals
Invezz.com - Grow your wealth with trading signalsInvezz.com - Grow your wealth with trading signals
Invezz.com - Grow your wealth with trading signals
 
Al Barsha Escorts $#$ O565212860 $#$ Escort Service In Al Barsha
Al Barsha Escorts $#$ O565212860 $#$ Escort Service In Al BarshaAl Barsha Escorts $#$ O565212860 $#$ Escort Service In Al Barsha
Al Barsha Escorts $#$ O565212860 $#$ Escort Service In Al Barsha
 
BigBuy dropshipping via API with DroFx.pptx
BigBuy dropshipping via API with DroFx.pptxBigBuy dropshipping via API with DroFx.pptx
BigBuy dropshipping via API with DroFx.pptx
 
Generative AI on Enterprise Cloud with NiFi and Milvus
Generative AI on Enterprise Cloud with NiFi and MilvusGenerative AI on Enterprise Cloud with NiFi and Milvus
Generative AI on Enterprise Cloud with NiFi and Milvus
 
BabyOno dropshipping via API with DroFx.pptx
BabyOno dropshipping via API with DroFx.pptxBabyOno dropshipping via API with DroFx.pptx
BabyOno dropshipping via API with DroFx.pptx
 
Midocean dropshipping via API with DroFx
Midocean dropshipping via API with DroFxMidocean dropshipping via API with DroFx
Midocean dropshipping via API with DroFx
 
Vip Model Call Girls (Delhi) Karol Bagh 9711199171✔️Body to body massage wit...
Vip Model  Call Girls (Delhi) Karol Bagh 9711199171✔️Body to body massage wit...Vip Model  Call Girls (Delhi) Karol Bagh 9711199171✔️Body to body massage wit...
Vip Model Call Girls (Delhi) Karol Bagh 9711199171✔️Body to body massage wit...
 
Ravak dropshipping via API with DroFx.pptx
Ravak dropshipping via API with DroFx.pptxRavak dropshipping via API with DroFx.pptx
Ravak dropshipping via API with DroFx.pptx
 
BDSM⚡Call Girls in Mandawali Delhi >༒8448380779 Escort Service
BDSM⚡Call Girls in Mandawali Delhi >༒8448380779 Escort ServiceBDSM⚡Call Girls in Mandawali Delhi >༒8448380779 Escort Service
BDSM⚡Call Girls in Mandawali Delhi >༒8448380779 Escort Service
 
Best VIP Call Girls Noida Sector 39 Call Me: 8448380779
Best VIP Call Girls Noida Sector 39 Call Me: 8448380779Best VIP Call Girls Noida Sector 39 Call Me: 8448380779
Best VIP Call Girls Noida Sector 39 Call Me: 8448380779
 
BPAC WITH UFSBI GENERAL PRESENTATION 18_05_2017-1.pptx
BPAC WITH UFSBI GENERAL PRESENTATION 18_05_2017-1.pptxBPAC WITH UFSBI GENERAL PRESENTATION 18_05_2017-1.pptx
BPAC WITH UFSBI GENERAL PRESENTATION 18_05_2017-1.pptx
 

Active Learning and Human-in-the-Loop

Notes de l'éditeur

  1. Over 200,000 Records
  2. 59,000 records
  3. Unlike humans, artificial intelligence has no ego, so it can make an unbiased estimate of its confidence - Where it’s confident we use its answer, because hardware CPUs get cheaper+faster every year and human CPUs don’t - Where it’s not confident we use a human because in real business applications 80% accuracy isn’t good enough
  4. I think we can and should apply this to every business process We start with a machine learning classifier. Unlike humans, artificial intelligence has no ego, so it can make an unbiased estimate of its confidence - Where it’s confident we use its answer, because hardware CPUs get cheaper+faster every year and human CPUs don’t - Where it’s not confident we use a human because in real business applications 80% accuracy isn’t good enough A huge side benefit is that the human labels can be reused used to improve the machine learning classifier over time. We didn’t invent any of this, lot’s of people are talking about this and thinking about this, including many people in the room. But looking at the industry we see a lot more people talking about it than actually doing it. We are going to make this setup so easy that you will have no excuse for not doing it.
  5. I think we can and should apply this to every business process We start with a machine learning classifier. Unlike humans, artificial intelligence has no ego, so it can make an unbiased estimate of its confidence - Where it’s confident we use its answer, because hardware CPUs get cheaper+faster every year and human CPUs don’t - Where it’s not confident we use a human because in real business applications 80% accuracy isn’t good enough A huge side benefit is that the human labels can be reused used to improve the machine learning classifier over time. We didn’t invent any of this, lot’s of people are talking about this and thinking about this, including many people in the room. But looking at the industry we see a lot more people talking about it than actually doing it. We are going to make this setup so easy that you will have no excuse for not doing it.
  6. A huge side benefit is that the human labels can be reused used to improve the machine learning classifier over time.
  7. handed control to the driver 272 times and a test driver felt compelled to intervene 69 times
  8. In the field of chess computers passed humans a long time ago. But if you really want to make a great chess playing algorithm you would still use a human and computer together. There is a subculture of folks who still play “Advanced Chess” and this is actually where the highest quality chess games take place. - Still situations where humans are better