Confidential1
James Orton | Australia and New Zealand | Data Scientist
AI and AutoML:
Debunking Myths
2020
Confidential2 Confidential2
AI and AutoML: Debunking Myths
Are they overhyped or the answer to our problems?
Beyond the hyperbole, what are autoML and AI?
How are they helpful, and when are they not?
Why are they more relevant and valuable than ever?
Our world is changing rapidly, many organisations will need to adapt quickly.
AI and AutoML are not magic but it can be transformative, find out how today!
Get practical tips and see AutoML in action with a real world example. We’ll
demonstrate how AutoML can augment your Data Scientists, supercharging your
team and giving your organisation the AI edge in record time.
Confidential3 Confidential3
James Orton
Data Scientist @ H2O.ai
Australia and New Zealand
Connect with me
https://www.linkedin.com/in/jamesortonthedataman/
james.orton@h2o.ai
… and who are H2O.ai?
Who is that talking?
Confidential4
Founded in Silicon Valley 2012
Funding: Series D
Investors: Goldman Sachs, Ping An,
Wells Fargo, NVIDIA, Nexus Ventures
We are Established
We Make World-class AI Platforms
We are Global
H2O Open Source Machine Learning
H2O Driverless AI: Automatic Machine Learning
H2O Q: AI platform for business users
Mountain View, NYC, London, Paris, Ottawa,
Prague, Chennai, Singapore, Melbourne
220+ 1K
20K 180K
Universities
Companies Using
H2O Open Source
Meetup Members
Experts
H2O.ai Snapshot
We are Passionate about Customers
Commonwealth Bank Australia, IP Australia,
Customer Service NSW, Aetna/CVS, Allergan,
AT&T, CapitalOne, Citi, Coca Cola, Bredesco, Dish,
Disney, Franklin Templeton, Genentech, Kaiser
Permanente, Lego, Merck, Pepsi, Reckitt Benckiser
Confidential5
Gartner 2020: H2O.ai is a Visionary in Two MQs
A new MQ and
the only AI
platform
company in the
quadrant.
2020 Cloud AI for Developer
Services MQ
2020 Data Science and Machine
Learning MQ
Named a Visionary,
with the strongest
“Completeness of
Vision” in the entire
quadrant.
Strengths:
1. Automation
2. Explainability
3. High-Performance ML Components
Strengths:
1. Automation
2. Ease of Use and Explainability
3. Excellent Customer Support
Confidential6 Confidential6
But what is AI?
Confidential7
What is Artificial Intelligence: Use cases, H2O.ai works on
with our customers:
Save Time. Save Money. Gain a Competitive Edge.
Wholesale / Commercial
Banking
• Know Your Customers (KYC)
• Anti-Money Laundering (AML)
Card / Payments Business
• Transaction frauds
• Collusion fraud
• Real-time targeting
• Credit risk scoring
• In-context promotion
Retail Banking
• Deposit fraud
• Customer churn prediction
• Auto-loan
Financial Services
• Early cancer detection
• Product recommendations
• Personalized prescription
matching
• Medical claim fraud detection
• Flu season prediction
• Drug discovery
• ER and hospital
management
• Remote patient monitoring
• Medical test predictions
Healthcare and
Life Science
• Predictive maintenance
• Avoidable truck-rolls
• Customer churn prediction
• Improved customer viewing
experience
• Master data management
• In-context promotions
• Intelligent ad placements
• Personalized program
recommendations
Telecom
• Funnel predictions
• Personalized ads
• Credit scoring
• Fraud detection
• Next best offer
• Next best action
• Customer segmentation
• Customer churn
• Customer recommendations
• Ad predictions and fraud
Marketing and Retail
Confidential8
Examples of the impact of AI Transformations
…real-time individualized experience
…dynamic yield optimizationBreak then fix
…personalized quality of serviceCustomer service silos
…personalized healthcareMass treatment
…real-time trade surveillanceDaily risk analysis
Mass branding
WITH AIPRE-AI
AI allows
organizations to
shift interactions
from…
Reactive
Post Transaction
Proactive
Pre Decision
Confidential9
How H2O.ai is Contributing to COVID-19
Expertise
H2O.ai’s data science experts
are contributing their
knowledge to solve pressing
problems with the pandemic
AI Platforms
H2O.ai is contributing its
Driverless AI and Q platform
to model, predict, and
visualize data sets
Sri Ambati
CEO and Founder, H2O.ai
1. Hospital staffing predictions
2. ICU transfers and triage
3. Population risk segmentation
4. Predicting the spread of COVID-19.
5. Predicting operational efficiency and
resilience during a pandemic
6. Hospital supply chain predictions
7. Predicting responses by city,
hospitals
8. Sepsis predictions
Problems we are solving
“
Data Sets
H2O.ai is evaluating global
and open health data sets to
determine patterns
“Data Science can save
lives today. AI is an
incredible force to do
good for humanity.”
AI Solutions
H2O.ai is creating pandemic
and health specific solutions
for general use
Confidential10 Confidential10
What is AutoML?
Automated machine learning (AutoML) is the process
of automating the process of applying machine learning to
real-world problems. AutoML covers the complete pipeline
from the raw dataset to the deployable machine learning
model. AutoML was proposed as an artificial intelligence-
based solution to the ever-growing challenge of applying
machine learning.
Confidential11
Prepare Data
ML Algorithms
Models
New Data
Prediction
Deployed Model
Explanation
Explanation
Features
(Original + Engineered)
Hyperparameters
AI App
End User
ML AlgorithmsML Algorithms
ModelsModels
Deployed
Model
Typical ML Workflow
Model
Explanation
Model
Report
Model
Management &
Monitoring
Model
Engineering
Tuning
(Scorer)
Training Data Explore Data
Confidential12 Confidential12
Not all AutoML is created equal
Confidential13 Confidential13
So we know what AI, ML and AutoML are…
We can see that AI and ML are already having a big
impact across many sectors of the economy
… but why use AutoML?
Confidential14 Confidential14
Why now?
Lets address
the elephant in
the room, or
why we are all
in separate
rooms?
Confidential15
After Covid-19 the World Will
Never Be the Same. But Maybe, It
Can Be Better!
1
5
Confidentia
l
Vanessa Bates Ramirez,
SingularityHub
Confidential16 Confidential16
Why now?
The model you built last year
to identify credit risk probably
is not relevant anymore
The world is changing at a
fast pace
How do you rapidly iterate,
build and deploy ML to keep
up?
Confidential17 Confidential17
Does not replace data scientists
Look for a solution that augments your current capability
Often can not do end to end data preparation
But it can tackle components very well, speak to us about auto data augmentation
Does not replace domain knowledge
Look for a solution that allows you to inject your own business specific knowledge
Does not, by itself, create an AI culture or data literacy
But it can create space for this
AutoML: What are some of the cautions?
Confidential18
Who is on the team?
Business leader, data
scientists, IT professional
Determine the problems you
want to solve with metrics (time,
money, # of customers, etc)
Determine where you have
data, need data, and can
use technology to find
answers and predictions.
Find answers efficiently.
Learn from others in the
data science community
Ask the Right Questions
Data & Technology
Community
Create a Data Culture
Understand and explain the
models. Use leading edge
technologies to guard for
bias, explain a model, and
present this to regulators
Trust in AI
2
1
3
4
5
5 Keys
to unlock your AI
Confidential19 Confidential19
A Very Simplistic AI project
Framing
• Culture and
Community
• The right
questions
Technical
Execution
• Do ML!
Impact
• Building
Trust
• Deployment
Confidential20 Confidential20
Manual ML
Framing
Technical Execution
Impact
Confidential21 Confidential21
Auto ML
Framing
Technical
Execution
Impact
Confidential22 Confidential22
Let’s see H2O.ai AutoML in action
Confidential23 Confidential23
• Automatic feature engineering,
machine learning and interpretability
• Fully automated machine learning
from ingest to deployment
• User licenses on a per seat basis
annually
• GUI-based interface for end-to-end
data science
• A new and innovated
platform to make your own
AI apps
• Enterprise commercial
software
• Easy and intuitive platform to
have AI answer your
question
H2O.ai: AI Platforms
In-memory, distributed
machine learning algorithms with
H2O Flow GUI
Open Source H2O Driverless AI H2O Q
• 100% open source – Apache
V2 Licensed
• Integration with Apache Spark
• Enterprise support subscriptions
• Interface using R, Python on
H2O Flow
H2O Model Ops
• AI deployment platform built
for DevOps and MLOps
• Scalable to support high
throughput and low latency
model scoring environments
• Comprehensive model
monitoring
• Drift Detection and retrain
ModelOps
Confidential24
The AI
Advantage
Using H2O
Driverless AI
Talent
Enterprise-ready
Get answers faster
Augment existing teams
• Develop AI models faster
• Saving Time
• Deploy production-ready models in hours vs. months
• New models with better accuracy
• Benchmarking existing models for better accuracy
• Allow data scientists of any level of expertise to develop
production-ready ML models
• Target more use cases
• Answer to increasing business demand to predict outcomes
• Get results across the business
Cross-enterprise efforts to scale AI
Enhance productivity
Improve models
Confidential25 Confidential25
Driverless AI
Features Targe
t
Data Quality and
Transformation
Modeling
Table
Model
Building
Model
Data Integration
+
Driverless AI:
Automates Data Science and ML Workflows
Highly Iterative Process
Confidential26
Challenges in AI Model Development
Basic Encoding
Feature Generation
Advanced Encoding
Feature Engineering
Algorithm Selection
Parameter Tuning
Model Building
Model Ensembles
Pipeline Generation
Model Explainabilty
Model Deployment
Model Documentation
• Time consuming
• Requires advanced
skill set
• Creating new feature
combinations requires
advanced skill
• Time consuming
• Requires advanced
knowledge of
algorithms and
parameters
• Creating ensembles
is an advanced skill
• Time consuming
• Requires different set of skills to
deploy models
• Explaining how models make
decisions is critical to building
trust with business stakeholders
and regulators
The entire process is highly iterative and can take weeks or months to develop a single
production-ready model.
Confidential27
H2O.ai sets an example in providing rich
explainability functionality, using diverse
techniques such as K-LIME, LIME-SUP,
Shapley, variable importance, decision tree
surrogate, ICE, partial dependence plots,
disparate impact analysis and “what-if
analysis.” The AutoDoc capability
automatically generates a complete set of
explanations in document format - Gartner
2020
Trust and Understanding of AI
Invited presentations: JSM (‘18, ‘19),
KDD (‘19); Accepted paper: NeurIPS
(‘19)
Confidential28 Confidential28
Catalog.h2o.ai (140+ BYOR Custom Recipes)
Confidential29
Confidential30
Automatic AI and ML
in a single platform
AI to do AI
Delivers insights
and interpretability
Customize and extend with
130+ open source recipes or
your domain expertise
30
Driverless AI: The Platform to Make Your Own AI
Confidential31
GET
STARTED
TODAY
• Learn about what healthcare, life sciences, finance and
insurance customers are doing with H2O.ai at our
website: https://www.h2o.ai/solutions/
• Take Driverless AI for a 21-day trial or
2 hour Free Cloud Test Drive and tutorials
• Meet the Makers at an event or meetup near you
• Watch a webinar to learn more
• Follow us on LinkedIn or Twitter @h2oai
Confidential32 Confidential32
Thank you
Questions?

AI and AutoML: Debunking Myths

  • 1.
    Confidential1 James Orton |Australia and New Zealand | Data Scientist AI and AutoML: Debunking Myths 2020
  • 2.
    Confidential2 Confidential2 AI andAutoML: Debunking Myths Are they overhyped or the answer to our problems? Beyond the hyperbole, what are autoML and AI? How are they helpful, and when are they not? Why are they more relevant and valuable than ever? Our world is changing rapidly, many organisations will need to adapt quickly. AI and AutoML are not magic but it can be transformative, find out how today! Get practical tips and see AutoML in action with a real world example. We’ll demonstrate how AutoML can augment your Data Scientists, supercharging your team and giving your organisation the AI edge in record time.
  • 3.
    Confidential3 Confidential3 James Orton DataScientist @ H2O.ai Australia and New Zealand Connect with me https://www.linkedin.com/in/jamesortonthedataman/ james.orton@h2o.ai … and who are H2O.ai? Who is that talking?
  • 4.
    Confidential4 Founded in SiliconValley 2012 Funding: Series D Investors: Goldman Sachs, Ping An, Wells Fargo, NVIDIA, Nexus Ventures We are Established We Make World-class AI Platforms We are Global H2O Open Source Machine Learning H2O Driverless AI: Automatic Machine Learning H2O Q: AI platform for business users Mountain View, NYC, London, Paris, Ottawa, Prague, Chennai, Singapore, Melbourne 220+ 1K 20K 180K Universities Companies Using H2O Open Source Meetup Members Experts H2O.ai Snapshot We are Passionate about Customers Commonwealth Bank Australia, IP Australia, Customer Service NSW, Aetna/CVS, Allergan, AT&T, CapitalOne, Citi, Coca Cola, Bredesco, Dish, Disney, Franklin Templeton, Genentech, Kaiser Permanente, Lego, Merck, Pepsi, Reckitt Benckiser
  • 5.
    Confidential5 Gartner 2020: H2O.aiis a Visionary in Two MQs A new MQ and the only AI platform company in the quadrant. 2020 Cloud AI for Developer Services MQ 2020 Data Science and Machine Learning MQ Named a Visionary, with the strongest “Completeness of Vision” in the entire quadrant. Strengths: 1. Automation 2. Explainability 3. High-Performance ML Components Strengths: 1. Automation 2. Ease of Use and Explainability 3. Excellent Customer Support
  • 6.
  • 7.
    Confidential7 What is ArtificialIntelligence: Use cases, H2O.ai works on with our customers: Save Time. Save Money. Gain a Competitive Edge. Wholesale / Commercial Banking • Know Your Customers (KYC) • Anti-Money Laundering (AML) Card / Payments Business • Transaction frauds • Collusion fraud • Real-time targeting • Credit risk scoring • In-context promotion Retail Banking • Deposit fraud • Customer churn prediction • Auto-loan Financial Services • Early cancer detection • Product recommendations • Personalized prescription matching • Medical claim fraud detection • Flu season prediction • Drug discovery • ER and hospital management • Remote patient monitoring • Medical test predictions Healthcare and Life Science • Predictive maintenance • Avoidable truck-rolls • Customer churn prediction • Improved customer viewing experience • Master data management • In-context promotions • Intelligent ad placements • Personalized program recommendations Telecom • Funnel predictions • Personalized ads • Credit scoring • Fraud detection • Next best offer • Next best action • Customer segmentation • Customer churn • Customer recommendations • Ad predictions and fraud Marketing and Retail
  • 8.
    Confidential8 Examples of theimpact of AI Transformations …real-time individualized experience …dynamic yield optimizationBreak then fix …personalized quality of serviceCustomer service silos …personalized healthcareMass treatment …real-time trade surveillanceDaily risk analysis Mass branding WITH AIPRE-AI AI allows organizations to shift interactions from… Reactive Post Transaction Proactive Pre Decision
  • 9.
    Confidential9 How H2O.ai isContributing to COVID-19 Expertise H2O.ai’s data science experts are contributing their knowledge to solve pressing problems with the pandemic AI Platforms H2O.ai is contributing its Driverless AI and Q platform to model, predict, and visualize data sets Sri Ambati CEO and Founder, H2O.ai 1. Hospital staffing predictions 2. ICU transfers and triage 3. Population risk segmentation 4. Predicting the spread of COVID-19. 5. Predicting operational efficiency and resilience during a pandemic 6. Hospital supply chain predictions 7. Predicting responses by city, hospitals 8. Sepsis predictions Problems we are solving “ Data Sets H2O.ai is evaluating global and open health data sets to determine patterns “Data Science can save lives today. AI is an incredible force to do good for humanity.” AI Solutions H2O.ai is creating pandemic and health specific solutions for general use
  • 10.
    Confidential10 Confidential10 What isAutoML? Automated machine learning (AutoML) is the process of automating the process of applying machine learning to real-world problems. AutoML covers the complete pipeline from the raw dataset to the deployable machine learning model. AutoML was proposed as an artificial intelligence- based solution to the ever-growing challenge of applying machine learning.
  • 11.
    Confidential11 Prepare Data ML Algorithms Models NewData Prediction Deployed Model Explanation Explanation Features (Original + Engineered) Hyperparameters AI App End User ML AlgorithmsML Algorithms ModelsModels Deployed Model Typical ML Workflow Model Explanation Model Report Model Management & Monitoring Model Engineering Tuning (Scorer) Training Data Explore Data
  • 12.
  • 13.
    Confidential13 Confidential13 So weknow what AI, ML and AutoML are… We can see that AI and ML are already having a big impact across many sectors of the economy … but why use AutoML?
  • 14.
    Confidential14 Confidential14 Why now? Letsaddress the elephant in the room, or why we are all in separate rooms?
  • 15.
    Confidential15 After Covid-19 theWorld Will Never Be the Same. But Maybe, It Can Be Better! 1 5 Confidentia l Vanessa Bates Ramirez, SingularityHub
  • 16.
    Confidential16 Confidential16 Why now? Themodel you built last year to identify credit risk probably is not relevant anymore The world is changing at a fast pace How do you rapidly iterate, build and deploy ML to keep up?
  • 17.
    Confidential17 Confidential17 Does notreplace data scientists Look for a solution that augments your current capability Often can not do end to end data preparation But it can tackle components very well, speak to us about auto data augmentation Does not replace domain knowledge Look for a solution that allows you to inject your own business specific knowledge Does not, by itself, create an AI culture or data literacy But it can create space for this AutoML: What are some of the cautions?
  • 18.
    Confidential18 Who is onthe team? Business leader, data scientists, IT professional Determine the problems you want to solve with metrics (time, money, # of customers, etc) Determine where you have data, need data, and can use technology to find answers and predictions. Find answers efficiently. Learn from others in the data science community Ask the Right Questions Data & Technology Community Create a Data Culture Understand and explain the models. Use leading edge technologies to guard for bias, explain a model, and present this to regulators Trust in AI 2 1 3 4 5 5 Keys to unlock your AI
  • 19.
    Confidential19 Confidential19 A VerySimplistic AI project Framing • Culture and Community • The right questions Technical Execution • Do ML! Impact • Building Trust • Deployment
  • 20.
  • 21.
  • 22.
  • 23.
    Confidential23 Confidential23 • Automaticfeature engineering, machine learning and interpretability • Fully automated machine learning from ingest to deployment • User licenses on a per seat basis annually • GUI-based interface for end-to-end data science • A new and innovated platform to make your own AI apps • Enterprise commercial software • Easy and intuitive platform to have AI answer your question H2O.ai: AI Platforms In-memory, distributed machine learning algorithms with H2O Flow GUI Open Source H2O Driverless AI H2O Q • 100% open source – Apache V2 Licensed • Integration with Apache Spark • Enterprise support subscriptions • Interface using R, Python on H2O Flow H2O Model Ops • AI deployment platform built for DevOps and MLOps • Scalable to support high throughput and low latency model scoring environments • Comprehensive model monitoring • Drift Detection and retrain ModelOps
  • 24.
    Confidential24 The AI Advantage Using H2O DriverlessAI Talent Enterprise-ready Get answers faster Augment existing teams • Develop AI models faster • Saving Time • Deploy production-ready models in hours vs. months • New models with better accuracy • Benchmarking existing models for better accuracy • Allow data scientists of any level of expertise to develop production-ready ML models • Target more use cases • Answer to increasing business demand to predict outcomes • Get results across the business Cross-enterprise efforts to scale AI Enhance productivity Improve models
  • 25.
    Confidential25 Confidential25 Driverless AI FeaturesTarge t Data Quality and Transformation Modeling Table Model Building Model Data Integration + Driverless AI: Automates Data Science and ML Workflows Highly Iterative Process
  • 26.
    Confidential26 Challenges in AIModel Development Basic Encoding Feature Generation Advanced Encoding Feature Engineering Algorithm Selection Parameter Tuning Model Building Model Ensembles Pipeline Generation Model Explainabilty Model Deployment Model Documentation • Time consuming • Requires advanced skill set • Creating new feature combinations requires advanced skill • Time consuming • Requires advanced knowledge of algorithms and parameters • Creating ensembles is an advanced skill • Time consuming • Requires different set of skills to deploy models • Explaining how models make decisions is critical to building trust with business stakeholders and regulators The entire process is highly iterative and can take weeks or months to develop a single production-ready model.
  • 27.
    Confidential27 H2O.ai sets anexample in providing rich explainability functionality, using diverse techniques such as K-LIME, LIME-SUP, Shapley, variable importance, decision tree surrogate, ICE, partial dependence plots, disparate impact analysis and “what-if analysis.” The AutoDoc capability automatically generates a complete set of explanations in document format - Gartner 2020 Trust and Understanding of AI Invited presentations: JSM (‘18, ‘19), KDD (‘19); Accepted paper: NeurIPS (‘19)
  • 28.
  • 29.
  • 30.
    Confidential30 Automatic AI andML in a single platform AI to do AI Delivers insights and interpretability Customize and extend with 130+ open source recipes or your domain expertise 30 Driverless AI: The Platform to Make Your Own AI
  • 31.
    Confidential31 GET STARTED TODAY • Learn aboutwhat healthcare, life sciences, finance and insurance customers are doing with H2O.ai at our website: https://www.h2o.ai/solutions/ • Take Driverless AI for a 21-day trial or 2 hour Free Cloud Test Drive and tutorials • Meet the Makers at an event or meetup near you • Watch a webinar to learn more • Follow us on LinkedIn or Twitter @h2oai
  • 32.
  • 33.