SlideShare uma empresa Scribd logo
1 de 30
AI: Intro and Use Cases
John T. Langton
June 27, 2019
Who am I?
 Ph.D. in computer science
 Director of Data Science at WK
 PI on multiple DoD projects
 Founder of VisiTrend: cyber ml
 Several peer reviewed publications and
speaking engagements
2
John T. Langton
Agenda
 Putting AI hype in perspective
 What is AI?
 An example of how AI works
 AI use cases
 Questions
3
AI: the hype
4
Bloomberg: number of AI mentions in corporate earnings statements
AI Hype Cycle
5
Gartner says ML is is
2-5 years to adoption
AI Reality
Machine learning has been in production for years, and it’s ubiquitous
 Siri in 2011
 Google search
 Netflix recommendations
 Image search
 Elevators
 Thermostats
 Rice cookers
AI from Adoption to Disruption 6
Multiple AI Hype Cycles: the story of AI Winters
As soon as AI solves something, but it’s not a walking,
talking robot, then it is no longer considered AI.
AI from Adoption to Disruption 7
AI from Adoption to Disruption 8
“We tend to overestimate the effect of
a technology in the short run and
underestimate the effect in the long
run”
– Roy Amara
 AI is not new; it’s ubiquitous and in products that you use every day
 What’s new are the application areas and scale of AI
 Digitization of content
 Advances in hardware
Why AI and Why Now?
Google’s TPU
9
So what is AI?
 A set of algorithms (search + statistics)
 Software that can program itself using data and/or axioms and a goal
 Can adapt over time
AI from Adoption to Disruption 10
The Birth of AI
Antiquity
1800’s
1947
1956
AI from Adoption to Disruption 11
Aristotle covers syllogism and goal-regression planning
Logical positivists: knowledge can be reasoned about mathematically
Alan Turing: the Turing Test, machine learning and genetic algorithms
McCarthy workshop at Dartmouth coined “artificial intelligence”
Most people think “Strong AI” but mean “Weak AI”
What Do We Mean by AI?
12
•Ability to reason like a human
•General problem solver
•Passes Turing Test (Blade Runner)
Strong or
General AI
•Well scoped applications such as predicting sepsis
•Automation, detection, prediction
•Data driven and often performs better than a human
Weak or
Narrow AI
General AI Requirements Turn into Narrow AI Subfields
 Natural Language Processing (NLP): topic modeling, sentiment analysis
 Perception: computer vision, speech recognition
 Knowledge representation: linked-lists, ontologies, frames, logic statements
 Reasoning: search, propositional and fuzzy logic, probabilistic reasoning
 Planning: decision theory, constraint satisfaction, problem solving as search
 Learning: machine learning, evolutionary computing
AI from Adoption to Disruption 13
What is AI
14
NLP
Machine Learning (ML)
Supervised Learning
Continuous
Target Variable
Regression
Forecasting
Drug Utilization
Categorical
Target Variable
Classification
Predicting
Sepsis
Unsupervised Learning
No Target Variable
Clustering
Customer
Segmentation
Association
Rules
Market Basket
Analysis
Reinforcement Learning
Fitness Functions
Genetic
Algorithms
Route
Optimization
Hidden Markov
Models
Spam Detection
Machine Learning is Pretty Big
AI from Adoption to Disruption 16
 Expert systems
 Discriminant analysis
 Agents / multi-agent architectures (“Society of Mind”)
 Evolutionary Algorithms
 Case Based Reasoning
 Fuzzy Logic
 Constraint-based satisfaction
 Game theory
 Search-based Problem Solving (A*, Hill Climbing, Local Beam)
 Pattern Recognition
 Sequence learning, classification, and optimization
 NLP: LDA, LSA, TF / IDF
 …
AI is Bigger than ML and Deep Learning
17
 Doctors – expert opinion from training, studies, dozens of cases
 Studies and Randomized Controlled Trials – up to hundreds of patients in each
 SIRS and qSOFA diagnosis systems – consensus of expert opinions + studies
 Over a million patients are diagnosed with sepsis every year
 OSU conducts study with 319,817 patients revealing flaws in SIRS and
qSOFA
 How do we leverage the scale of this data?!
AI Example: predicting / detecting sepsis
1,000,000
Gini = .5
p(s) = .5
p(s) = .7
Gini = .58 = 1-(.72 +.32)
714,000
p(s) = 0
Gini = 0 = 1-(12 +02)
286,000
AI Example: predicting Sepsis with CART
PCT ≤ .2 ng/mL truefalse
For each variable:
 Compute data distribution
 Find optimal split point(s)
 Compute Gini index for each branch
 Sum Gini indexes weighted by # of
records for each branch
1,000,000
Gini = .5
p(s) = .5
p(s) = .7
Gini = .58 = 1-(.72 +.32)
714,000
p(s) = .81
Gini = .69 = 1-(.812 +.192)
614,000
p(s) = 1
Gini = 0 = 1-(12 +02)
100,000
p(s) = 0
Gini = 0 = 1-(12 +02)
286,000
AI Example: predicting Sepsis with CART
• Repeat for each variable
• Choose variable with lowest Gini sum and split
• Repeat until stopping criteria
• Branch has Gini = 0
• No more variables
• Maximum depth
PCT ≤0.2 ng/mL
systolic ≤ 90 mmHg
truefalse
false true
 Tree pruning
 Other cost functions
 Winnowing variables using dimensionality
reduction (PCA, MDS)
 Weighting classes and misclassification types
 Sampling techniques for data points and
variables (stratified sampling, bootstrapping)
 Ensemble methods like boosting and rf
 # of trees
 Tree depth
 Learning rate
AI Example: more advanced methods
AI Evaluation Example: 10-fold Cross Validation
AI: Metrics of Evaluation
AI Use Cases
 Cybersecurity
 Anomaly detection
 Static analysis of malware
 Domain generating algorithm detection
 Tax and Legal
 Regulatory change management
 Billing automation and reconciliation
 Finance
 Risk analysis
 Projections and forecasting
24
Health AI Startups
AI from Adoption to Disruption 25
 Risk Analytics: readmission,
interventions, quality measures
 Medical Imaging: diabetic
retinopathy, skin cancer, radiology
 Text Analytics: terminology
mapping, code extraction,
document classification
 Predictive Analytics: Clostridium
Difficile, Sepsis, renal failure
Source: www.cbinsights.com
AI from Adoption to Disruption 26
 Companies:
 Freenome, SOPHiA Genetics, Deep Genomics, FDNA, DeCODE, Verily Life
Sciences (spun out of Google)
 Precision medicine: gene therapy development and biomarker discovery
 Drug discovery: predicting response to molecular compounds
 Risk analytics: detecting genetic variants associated with disease
AI + Genomics
AI and Radiology
AI from Adoption to Disruption 27
AI and Radiology
AI from Adoption to Disruption 28
FDA approves IDx-DR, AI-powered
software to detect diabetic retinopathy
Infervision uses AI to detect bleed
volume in stroke patients
Concerns, Obstacles, and Challenges
 Data
 Is there enough?
 Does it need to be labeled?
 Is it noisy?
 Class imbalanced?
 Black box vs explainable AI
 AI bias
 Regulations
 Evaluation (should AI drivers have less accidents than humans?)
AI from Adoption to Disruption 29
Questions?
 John T. Langton
 http://JohnLangton.com
AI from Adoption to Disruption 30

Mais conteúdo relacionado

Semelhante a What is AI

Artificial Intelligence - Potential Game Changer for Medical Technology Compa...
Artificial Intelligence - Potential Game Changer for Medical Technology Compa...Artificial Intelligence - Potential Game Changer for Medical Technology Compa...
Artificial Intelligence - Potential Game Changer for Medical Technology Compa...CitiusTech
 
Artificial intelligence in health care by Islam salama " Saimo#BoOm "
Artificial intelligence in health care by Islam salama " Saimo#BoOm "Artificial intelligence in health care by Islam salama " Saimo#BoOm "
Artificial intelligence in health care by Islam salama " Saimo#BoOm "Dr-Islam Salama
 
Frankie Rybicki slide set for Deep Learning in Radiology / Medicine
Frankie Rybicki slide set for Deep Learning in Radiology / MedicineFrankie Rybicki slide set for Deep Learning in Radiology / Medicine
Frankie Rybicki slide set for Deep Learning in Radiology / MedicineFrank Rybicki
 
Semantic, Cognitive and Perceptual Computing -Keynote artificial intelligence...
Semantic, Cognitive and Perceptual Computing -Keynote artificial intelligence...Semantic, Cognitive and Perceptual Computing -Keynote artificial intelligence...
Semantic, Cognitive and Perceptual Computing -Keynote artificial intelligence...Artificial Intelligence Institute at UofSC
 
Big Data and Artificial Intelligence in Critical Care
Big Data and Artificial Intelligence in Critical CareBig Data and Artificial Intelligence in Critical Care
Big Data and Artificial Intelligence in Critical CareTommaso Scquizzato
 
Artificial Intelligence: Role In Pharma Sector
Artificial Intelligence: Role In Pharma SectorArtificial Intelligence: Role In Pharma Sector
Artificial Intelligence: Role In Pharma SectorJayBhavsar41
 
introduction to machin learning
introduction to machin learningintroduction to machin learning
introduction to machin learningnilimapatel6
 
AI and covid19 | Mr. R. Rajkumar, Assistant Professor, Department of CSE
AI and covid19 | Mr. R. Rajkumar, Assistant Professor, Department of CSEAI and covid19 | Mr. R. Rajkumar, Assistant Professor, Department of CSE
AI and covid19 | Mr. R. Rajkumar, Assistant Professor, Department of CSERajkumar R
 
Dark Side and Bright Side of AI in Medicine
Dark Side and Bright Side of AI in MedicineDark Side and Bright Side of AI in Medicine
Dark Side and Bright Side of AI in MedicineUniversity of Florida
 
Overview of Data Science and AI
Overview of Data Science and AIOverview of Data Science and AI
Overview of Data Science and AIjohnstamford
 
IFMIA 2019 Plenary Talk : Deep Learning in Medicine; Engineers' Perspectives
IFMIA 2019 Plenary Talk : Deep Learning in Medicine; Engineers' PerspectivesIFMIA 2019 Plenary Talk : Deep Learning in Medicine; Engineers' Perspectives
IFMIA 2019 Plenary Talk : Deep Learning in Medicine; Engineers' PerspectivesNamkug Kim
 
Prediction, Big Data, and AI: Steyerberg, Basel Nov 1, 2019
Prediction, Big Data, and AI: Steyerberg, Basel Nov 1, 2019Prediction, Big Data, and AI: Steyerberg, Basel Nov 1, 2019
Prediction, Big Data, and AI: Steyerberg, Basel Nov 1, 2019Ewout Steyerberg
 
AI in Healthcare: Can AI Help in Diagnosing Coronavirus
AI in Healthcare: Can AI Help in Diagnosing CoronavirusAI in Healthcare: Can AI Help in Diagnosing Coronavirus
AI in Healthcare: Can AI Help in Diagnosing CoronavirusSkyl.ai
 
Trusted, Transparent and Fair AI using Open Source
Trusted, Transparent and Fair AI using Open SourceTrusted, Transparent and Fair AI using Open Source
Trusted, Transparent and Fair AI using Open SourceAnimesh Singh
 
New Developments in Machine Learning - Prof. Dr. Max Welling
New Developments in Machine Learning - Prof. Dr. Max WellingNew Developments in Machine Learning - Prof. Dr. Max Welling
New Developments in Machine Learning - Prof. Dr. Max WellingTextkernel
 
Machine learning in medicine: calm down
Machine learning in medicine: calm downMachine learning in medicine: calm down
Machine learning in medicine: calm downBenVanCalster
 
Inauguration Function - Ohio Center of Excellence in Knowledge-Enabled Comput...
Inauguration Function - Ohio Center of Excellence in Knowledge-Enabled Comput...Inauguration Function - Ohio Center of Excellence in Knowledge-Enabled Comput...
Inauguration Function - Ohio Center of Excellence in Knowledge-Enabled Comput...Artificial Intelligence Institute at UofSC
 

Semelhante a What is AI (20)

Artificial Intelligence - Potential Game Changer for Medical Technology Compa...
Artificial Intelligence - Potential Game Changer for Medical Technology Compa...Artificial Intelligence - Potential Game Changer for Medical Technology Compa...
Artificial Intelligence - Potential Game Changer for Medical Technology Compa...
 
Artificial intelligence in health care by Islam salama " Saimo#BoOm "
Artificial intelligence in health care by Islam salama " Saimo#BoOm "Artificial intelligence in health care by Islam salama " Saimo#BoOm "
Artificial intelligence in health care by Islam salama " Saimo#BoOm "
 
Frankie Rybicki slide set for Deep Learning in Radiology / Medicine
Frankie Rybicki slide set for Deep Learning in Radiology / MedicineFrankie Rybicki slide set for Deep Learning in Radiology / Medicine
Frankie Rybicki slide set for Deep Learning in Radiology / Medicine
 
Semantic, Cognitive and Perceptual Computing -Keynote artificial intelligence...
Semantic, Cognitive and Perceptual Computing -Keynote artificial intelligence...Semantic, Cognitive and Perceptual Computing -Keynote artificial intelligence...
Semantic, Cognitive and Perceptual Computing -Keynote artificial intelligence...
 
Big Data and Artificial Intelligence in Critical Care
Big Data and Artificial Intelligence in Critical CareBig Data and Artificial Intelligence in Critical Care
Big Data and Artificial Intelligence in Critical Care
 
Artificial Intelligence: Role In Pharma Sector
Artificial Intelligence: Role In Pharma SectorArtificial Intelligence: Role In Pharma Sector
Artificial Intelligence: Role In Pharma Sector
 
introduction to machin learning
introduction to machin learningintroduction to machin learning
introduction to machin learning
 
i2ml3e-chap1.pptx
i2ml3e-chap1.pptxi2ml3e-chap1.pptx
i2ml3e-chap1.pptx
 
AI and covid19 | Mr. R. Rajkumar, Assistant Professor, Department of CSE
AI and covid19 | Mr. R. Rajkumar, Assistant Professor, Department of CSEAI and covid19 | Mr. R. Rajkumar, Assistant Professor, Department of CSE
AI and covid19 | Mr. R. Rajkumar, Assistant Professor, Department of CSE
 
Dark Side and Bright Side of AI in Medicine
Dark Side and Bright Side of AI in MedicineDark Side and Bright Side of AI in Medicine
Dark Side and Bright Side of AI in Medicine
 
Overview of Data Science and AI
Overview of Data Science and AIOverview of Data Science and AI
Overview of Data Science and AI
 
AI in healthcare.pptx
AI in healthcare.pptxAI in healthcare.pptx
AI in healthcare.pptx
 
IFMIA 2019 Plenary Talk : Deep Learning in Medicine; Engineers' Perspectives
IFMIA 2019 Plenary Talk : Deep Learning in Medicine; Engineers' PerspectivesIFMIA 2019 Plenary Talk : Deep Learning in Medicine; Engineers' Perspectives
IFMIA 2019 Plenary Talk : Deep Learning in Medicine; Engineers' Perspectives
 
Prediction, Big Data, and AI: Steyerberg, Basel Nov 1, 2019
Prediction, Big Data, and AI: Steyerberg, Basel Nov 1, 2019Prediction, Big Data, and AI: Steyerberg, Basel Nov 1, 2019
Prediction, Big Data, and AI: Steyerberg, Basel Nov 1, 2019
 
AI in Healthcare: Can AI Help in Diagnosing Coronavirus
AI in Healthcare: Can AI Help in Diagnosing CoronavirusAI in Healthcare: Can AI Help in Diagnosing Coronavirus
AI in Healthcare: Can AI Help in Diagnosing Coronavirus
 
Lec 01
Lec 01Lec 01
Lec 01
 
Trusted, Transparent and Fair AI using Open Source
Trusted, Transparent and Fair AI using Open SourceTrusted, Transparent and Fair AI using Open Source
Trusted, Transparent and Fair AI using Open Source
 
New Developments in Machine Learning - Prof. Dr. Max Welling
New Developments in Machine Learning - Prof. Dr. Max WellingNew Developments in Machine Learning - Prof. Dr. Max Welling
New Developments in Machine Learning - Prof. Dr. Max Welling
 
Machine learning in medicine: calm down
Machine learning in medicine: calm downMachine learning in medicine: calm down
Machine learning in medicine: calm down
 
Inauguration Function - Ohio Center of Excellence in Knowledge-Enabled Comput...
Inauguration Function - Ohio Center of Excellence in Knowledge-Enabled Comput...Inauguration Function - Ohio Center of Excellence in Knowledge-Enabled Comput...
Inauguration Function - Ohio Center of Excellence in Knowledge-Enabled Comput...
 

Último

Principles and Practices of Data Visualization
Principles and Practices of Data VisualizationPrinciples and Practices of Data Visualization
Principles and Practices of Data VisualizationKianJazayeri1
 
Student Profile Sample report on improving academic performance by uniting gr...
Student Profile Sample report on improving academic performance by uniting gr...Student Profile Sample report on improving academic performance by uniting gr...
Student Profile Sample report on improving academic performance by uniting gr...Seán Kennedy
 
SMOTE and K-Fold Cross Validation-Presentation.pptx
SMOTE and K-Fold Cross Validation-Presentation.pptxSMOTE and K-Fold Cross Validation-Presentation.pptx
SMOTE and K-Fold Cross Validation-Presentation.pptxHaritikaChhatwal1
 
Data Analysis Project Presentation: Unveiling Your Ideal Customer, Bank Custo...
Data Analysis Project Presentation: Unveiling Your Ideal Customer, Bank Custo...Data Analysis Project Presentation: Unveiling Your Ideal Customer, Bank Custo...
Data Analysis Project Presentation: Unveiling Your Ideal Customer, Bank Custo...Boston Institute of Analytics
 
modul pembelajaran robotic Workshop _ by Slidesgo.pptx
modul pembelajaran robotic Workshop _ by Slidesgo.pptxmodul pembelajaran robotic Workshop _ by Slidesgo.pptx
modul pembelajaran robotic Workshop _ by Slidesgo.pptxaleedritatuxx
 
Data Factory in Microsoft Fabric (MsBIP #82)
Data Factory in Microsoft Fabric (MsBIP #82)Data Factory in Microsoft Fabric (MsBIP #82)
Data Factory in Microsoft Fabric (MsBIP #82)Cathrine Wilhelmsen
 
Semantic Shed - Squashing and Squeezing.pptx
Semantic Shed - Squashing and Squeezing.pptxSemantic Shed - Squashing and Squeezing.pptx
Semantic Shed - Squashing and Squeezing.pptxMike Bennett
 
Advanced Machine Learning for Business Professionals
Advanced Machine Learning for Business ProfessionalsAdvanced Machine Learning for Business Professionals
Advanced Machine Learning for Business ProfessionalsVICTOR MAESTRE RAMIREZ
 
Minimizing AI Hallucinations/Confabulations and the Path towards AGI with Exa...
Minimizing AI Hallucinations/Confabulations and the Path towards AGI with Exa...Minimizing AI Hallucinations/Confabulations and the Path towards AGI with Exa...
Minimizing AI Hallucinations/Confabulations and the Path towards AGI with Exa...Thomas Poetter
 
wepik-insightful-infographics-a-data-visualization-overview-20240401133220kwr...
wepik-insightful-infographics-a-data-visualization-overview-20240401133220kwr...wepik-insightful-infographics-a-data-visualization-overview-20240401133220kwr...
wepik-insightful-infographics-a-data-visualization-overview-20240401133220kwr...KarteekMane1
 
Learn How Data Science Changes Our World
Learn How Data Science Changes Our WorldLearn How Data Science Changes Our World
Learn How Data Science Changes Our WorldEduminds Learning
 
English-8-Q4-W3-Synthesizing-Essential-Information-From-Various-Sources-1.pdf
English-8-Q4-W3-Synthesizing-Essential-Information-From-Various-Sources-1.pdfEnglish-8-Q4-W3-Synthesizing-Essential-Information-From-Various-Sources-1.pdf
English-8-Q4-W3-Synthesizing-Essential-Information-From-Various-Sources-1.pdfblazblazml
 
Real-Time AI Streaming - AI Max Princeton
Real-Time AI  Streaming - AI Max PrincetonReal-Time AI  Streaming - AI Max Princeton
Real-Time AI Streaming - AI Max PrincetonTimothy Spann
 
6 Tips for Interpretable Topic Models _ by Nicha Ruchirawat _ Towards Data Sc...
6 Tips for Interpretable Topic Models _ by Nicha Ruchirawat _ Towards Data Sc...6 Tips for Interpretable Topic Models _ by Nicha Ruchirawat _ Towards Data Sc...
6 Tips for Interpretable Topic Models _ by Nicha Ruchirawat _ Towards Data Sc...Dr Arash Najmaei ( Phd., MBA, BSc)
 
Digital Marketing Plan, how digital marketing works
Digital Marketing Plan, how digital marketing worksDigital Marketing Plan, how digital marketing works
Digital Marketing Plan, how digital marketing worksdeepakthakur548787
 
Bank Loan Approval Analysis: A Comprehensive Data Analysis Project
Bank Loan Approval Analysis: A Comprehensive Data Analysis ProjectBank Loan Approval Analysis: A Comprehensive Data Analysis Project
Bank Loan Approval Analysis: A Comprehensive Data Analysis ProjectBoston Institute of Analytics
 
INTRODUCTION TO Natural language processing
INTRODUCTION TO Natural language processingINTRODUCTION TO Natural language processing
INTRODUCTION TO Natural language processingsocarem879
 
Cyber awareness ppt on the recorded data
Cyber awareness ppt on the recorded dataCyber awareness ppt on the recorded data
Cyber awareness ppt on the recorded dataTecnoIncentive
 

Último (20)

Principles and Practices of Data Visualization
Principles and Practices of Data VisualizationPrinciples and Practices of Data Visualization
Principles and Practices of Data Visualization
 
Student Profile Sample report on improving academic performance by uniting gr...
Student Profile Sample report on improving academic performance by uniting gr...Student Profile Sample report on improving academic performance by uniting gr...
Student Profile Sample report on improving academic performance by uniting gr...
 
SMOTE and K-Fold Cross Validation-Presentation.pptx
SMOTE and K-Fold Cross Validation-Presentation.pptxSMOTE and K-Fold Cross Validation-Presentation.pptx
SMOTE and K-Fold Cross Validation-Presentation.pptx
 
Data Analysis Project Presentation: Unveiling Your Ideal Customer, Bank Custo...
Data Analysis Project Presentation: Unveiling Your Ideal Customer, Bank Custo...Data Analysis Project Presentation: Unveiling Your Ideal Customer, Bank Custo...
Data Analysis Project Presentation: Unveiling Your Ideal Customer, Bank Custo...
 
modul pembelajaran robotic Workshop _ by Slidesgo.pptx
modul pembelajaran robotic Workshop _ by Slidesgo.pptxmodul pembelajaran robotic Workshop _ by Slidesgo.pptx
modul pembelajaran robotic Workshop _ by Slidesgo.pptx
 
Data Factory in Microsoft Fabric (MsBIP #82)
Data Factory in Microsoft Fabric (MsBIP #82)Data Factory in Microsoft Fabric (MsBIP #82)
Data Factory in Microsoft Fabric (MsBIP #82)
 
Semantic Shed - Squashing and Squeezing.pptx
Semantic Shed - Squashing and Squeezing.pptxSemantic Shed - Squashing and Squeezing.pptx
Semantic Shed - Squashing and Squeezing.pptx
 
Data Analysis Project: Stroke Prediction
Data Analysis Project: Stroke PredictionData Analysis Project: Stroke Prediction
Data Analysis Project: Stroke Prediction
 
Advanced Machine Learning for Business Professionals
Advanced Machine Learning for Business ProfessionalsAdvanced Machine Learning for Business Professionals
Advanced Machine Learning for Business Professionals
 
Minimizing AI Hallucinations/Confabulations and the Path towards AGI with Exa...
Minimizing AI Hallucinations/Confabulations and the Path towards AGI with Exa...Minimizing AI Hallucinations/Confabulations and the Path towards AGI with Exa...
Minimizing AI Hallucinations/Confabulations and the Path towards AGI with Exa...
 
wepik-insightful-infographics-a-data-visualization-overview-20240401133220kwr...
wepik-insightful-infographics-a-data-visualization-overview-20240401133220kwr...wepik-insightful-infographics-a-data-visualization-overview-20240401133220kwr...
wepik-insightful-infographics-a-data-visualization-overview-20240401133220kwr...
 
Learn How Data Science Changes Our World
Learn How Data Science Changes Our WorldLearn How Data Science Changes Our World
Learn How Data Science Changes Our World
 
English-8-Q4-W3-Synthesizing-Essential-Information-From-Various-Sources-1.pdf
English-8-Q4-W3-Synthesizing-Essential-Information-From-Various-Sources-1.pdfEnglish-8-Q4-W3-Synthesizing-Essential-Information-From-Various-Sources-1.pdf
English-8-Q4-W3-Synthesizing-Essential-Information-From-Various-Sources-1.pdf
 
Real-Time AI Streaming - AI Max Princeton
Real-Time AI  Streaming - AI Max PrincetonReal-Time AI  Streaming - AI Max Princeton
Real-Time AI Streaming - AI Max Princeton
 
6 Tips for Interpretable Topic Models _ by Nicha Ruchirawat _ Towards Data Sc...
6 Tips for Interpretable Topic Models _ by Nicha Ruchirawat _ Towards Data Sc...6 Tips for Interpretable Topic Models _ by Nicha Ruchirawat _ Towards Data Sc...
6 Tips for Interpretable Topic Models _ by Nicha Ruchirawat _ Towards Data Sc...
 
Insurance Churn Prediction Data Analysis Project
Insurance Churn Prediction Data Analysis ProjectInsurance Churn Prediction Data Analysis Project
Insurance Churn Prediction Data Analysis Project
 
Digital Marketing Plan, how digital marketing works
Digital Marketing Plan, how digital marketing worksDigital Marketing Plan, how digital marketing works
Digital Marketing Plan, how digital marketing works
 
Bank Loan Approval Analysis: A Comprehensive Data Analysis Project
Bank Loan Approval Analysis: A Comprehensive Data Analysis ProjectBank Loan Approval Analysis: A Comprehensive Data Analysis Project
Bank Loan Approval Analysis: A Comprehensive Data Analysis Project
 
INTRODUCTION TO Natural language processing
INTRODUCTION TO Natural language processingINTRODUCTION TO Natural language processing
INTRODUCTION TO Natural language processing
 
Cyber awareness ppt on the recorded data
Cyber awareness ppt on the recorded dataCyber awareness ppt on the recorded data
Cyber awareness ppt on the recorded data
 

What is AI

  • 1. AI: Intro and Use Cases John T. Langton June 27, 2019
  • 2. Who am I?  Ph.D. in computer science  Director of Data Science at WK  PI on multiple DoD projects  Founder of VisiTrend: cyber ml  Several peer reviewed publications and speaking engagements 2 John T. Langton
  • 3. Agenda  Putting AI hype in perspective  What is AI?  An example of how AI works  AI use cases  Questions 3
  • 4. AI: the hype 4 Bloomberg: number of AI mentions in corporate earnings statements
  • 5. AI Hype Cycle 5 Gartner says ML is is 2-5 years to adoption
  • 6. AI Reality Machine learning has been in production for years, and it’s ubiquitous  Siri in 2011  Google search  Netflix recommendations  Image search  Elevators  Thermostats  Rice cookers AI from Adoption to Disruption 6
  • 7. Multiple AI Hype Cycles: the story of AI Winters As soon as AI solves something, but it’s not a walking, talking robot, then it is no longer considered AI. AI from Adoption to Disruption 7
  • 8. AI from Adoption to Disruption 8 “We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run” – Roy Amara
  • 9.  AI is not new; it’s ubiquitous and in products that you use every day  What’s new are the application areas and scale of AI  Digitization of content  Advances in hardware Why AI and Why Now? Google’s TPU 9
  • 10. So what is AI?  A set of algorithms (search + statistics)  Software that can program itself using data and/or axioms and a goal  Can adapt over time AI from Adoption to Disruption 10
  • 11. The Birth of AI Antiquity 1800’s 1947 1956 AI from Adoption to Disruption 11 Aristotle covers syllogism and goal-regression planning Logical positivists: knowledge can be reasoned about mathematically Alan Turing: the Turing Test, machine learning and genetic algorithms McCarthy workshop at Dartmouth coined “artificial intelligence”
  • 12. Most people think “Strong AI” but mean “Weak AI” What Do We Mean by AI? 12 •Ability to reason like a human •General problem solver •Passes Turing Test (Blade Runner) Strong or General AI •Well scoped applications such as predicting sepsis •Automation, detection, prediction •Data driven and often performs better than a human Weak or Narrow AI
  • 13. General AI Requirements Turn into Narrow AI Subfields  Natural Language Processing (NLP): topic modeling, sentiment analysis  Perception: computer vision, speech recognition  Knowledge representation: linked-lists, ontologies, frames, logic statements  Reasoning: search, propositional and fuzzy logic, probabilistic reasoning  Planning: decision theory, constraint satisfaction, problem solving as search  Learning: machine learning, evolutionary computing AI from Adoption to Disruption 13
  • 15. Machine Learning (ML) Supervised Learning Continuous Target Variable Regression Forecasting Drug Utilization Categorical Target Variable Classification Predicting Sepsis Unsupervised Learning No Target Variable Clustering Customer Segmentation Association Rules Market Basket Analysis Reinforcement Learning Fitness Functions Genetic Algorithms Route Optimization Hidden Markov Models Spam Detection
  • 16. Machine Learning is Pretty Big AI from Adoption to Disruption 16
  • 17.  Expert systems  Discriminant analysis  Agents / multi-agent architectures (“Society of Mind”)  Evolutionary Algorithms  Case Based Reasoning  Fuzzy Logic  Constraint-based satisfaction  Game theory  Search-based Problem Solving (A*, Hill Climbing, Local Beam)  Pattern Recognition  Sequence learning, classification, and optimization  NLP: LDA, LSA, TF / IDF  … AI is Bigger than ML and Deep Learning 17
  • 18.  Doctors – expert opinion from training, studies, dozens of cases  Studies and Randomized Controlled Trials – up to hundreds of patients in each  SIRS and qSOFA diagnosis systems – consensus of expert opinions + studies  Over a million patients are diagnosed with sepsis every year  OSU conducts study with 319,817 patients revealing flaws in SIRS and qSOFA  How do we leverage the scale of this data?! AI Example: predicting / detecting sepsis
  • 19. 1,000,000 Gini = .5 p(s) = .5 p(s) = .7 Gini = .58 = 1-(.72 +.32) 714,000 p(s) = 0 Gini = 0 = 1-(12 +02) 286,000 AI Example: predicting Sepsis with CART PCT ≤ .2 ng/mL truefalse For each variable:  Compute data distribution  Find optimal split point(s)  Compute Gini index for each branch  Sum Gini indexes weighted by # of records for each branch
  • 20. 1,000,000 Gini = .5 p(s) = .5 p(s) = .7 Gini = .58 = 1-(.72 +.32) 714,000 p(s) = .81 Gini = .69 = 1-(.812 +.192) 614,000 p(s) = 1 Gini = 0 = 1-(12 +02) 100,000 p(s) = 0 Gini = 0 = 1-(12 +02) 286,000 AI Example: predicting Sepsis with CART • Repeat for each variable • Choose variable with lowest Gini sum and split • Repeat until stopping criteria • Branch has Gini = 0 • No more variables • Maximum depth PCT ≤0.2 ng/mL systolic ≤ 90 mmHg truefalse false true
  • 21.  Tree pruning  Other cost functions  Winnowing variables using dimensionality reduction (PCA, MDS)  Weighting classes and misclassification types  Sampling techniques for data points and variables (stratified sampling, bootstrapping)  Ensemble methods like boosting and rf  # of trees  Tree depth  Learning rate AI Example: more advanced methods
  • 22. AI Evaluation Example: 10-fold Cross Validation
  • 23. AI: Metrics of Evaluation
  • 24. AI Use Cases  Cybersecurity  Anomaly detection  Static analysis of malware  Domain generating algorithm detection  Tax and Legal  Regulatory change management  Billing automation and reconciliation  Finance  Risk analysis  Projections and forecasting 24
  • 25. Health AI Startups AI from Adoption to Disruption 25  Risk Analytics: readmission, interventions, quality measures  Medical Imaging: diabetic retinopathy, skin cancer, radiology  Text Analytics: terminology mapping, code extraction, document classification  Predictive Analytics: Clostridium Difficile, Sepsis, renal failure Source: www.cbinsights.com
  • 26. AI from Adoption to Disruption 26  Companies:  Freenome, SOPHiA Genetics, Deep Genomics, FDNA, DeCODE, Verily Life Sciences (spun out of Google)  Precision medicine: gene therapy development and biomarker discovery  Drug discovery: predicting response to molecular compounds  Risk analytics: detecting genetic variants associated with disease AI + Genomics
  • 27. AI and Radiology AI from Adoption to Disruption 27
  • 28. AI and Radiology AI from Adoption to Disruption 28 FDA approves IDx-DR, AI-powered software to detect diabetic retinopathy Infervision uses AI to detect bleed volume in stroke patients
  • 29. Concerns, Obstacles, and Challenges  Data  Is there enough?  Does it need to be labeled?  Is it noisy?  Class imbalanced?  Black box vs explainable AI  AI bias  Regulations  Evaluation (should AI drivers have less accidents than humans?) AI from Adoption to Disruption 29
  • 30. Questions?  John T. Langton  http://JohnLangton.com AI from Adoption to Disruption 30