strata - putting ai to work for business - it's a journey

35
Putting AI to work for business It's a Journey Strata 2018 - London Carlo Appugliese Program Director, Machine Learning IBM Analytics, Data Science Elite Team

Upload: others

Post on 23-Apr-2022

4 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Strata - Putting AI to work for business - It's a Journey

Putting AI to work for businessIt's a JourneyStrata 2018 - London

Carlo AppuglieseProgram Director, Machine Learning IBM Analytics, Data Science Elite Team

Page 2: Strata - Putting AI to work for business - It's a Journey

IBM Cloud / DOC ID / Month XX, 2017 / © 2017 IBM Corporation 2

Artificial Intelligence…

Has Sci-Fi become a reality??? Maybe Soon!!

Page 3: Strata - Putting AI to work for business - It's a Journey

IBM Cloud / DOC ID / Month XX, 2017 / © 2017 IBM Corporation 3

Artificial Intelligence, What has it done for you..

Personalized Content…

And of course games….

Avoid Humans…

Page 4: Strata - Putting AI to work for business - It's a Journey

IBM Cloud / DOC ID / Month XX, 2017 / © 2017 IBM Corporation 4

Artificial Intelligence, for Business..

Personalized Content…Netflix…. 1 billion….“With better search results, Netflix estimated that it avoids 1 billion annually in cancelled subscriptions.” Blizzard… 8 Billion Annually…

Avoid Humans…Priceless!!!!

Page 5: Strata - Putting AI to work for business - It's a Journey

Artificial Intelligence…

“Over the next decade, AI won’t replace managers, but managers who use AI will replace those who don’t.”

“Business of Artificial Intelligence” by Harvard Business ReviewSource: https://hbr.org/cover-story/2017/07/the-business-of-artificial-intelligence

Page 6: Strata - Putting AI to work for business - It's a Journey

IBM Cloud / DOC ID / Month XX, 2017 / © 2017 IBM Corporation 6

Artificial Intelligence, impact on business…

A single use case in large US based Insurance Company,Built a simple ML model that cost about 150K, But in production it saved the company 15 Million annually..

“AI means big $$$$ for Business…..” - Carlo

“We cannot solve our problems with the same thinking we used when we created them.” – Albert Einstein

Page 7: Strata - Putting AI to work for business - It's a Journey

Understanding the Hype.. Gartner

7Think 2018 / DOC ID / Month XX, 2018 / © 2018 IBM Corporation

“We always overestimate the change that will occur in the next two years and underestimate the change that will occur in the next ten. Don't let yourself be lulled into inaction.” Bill Gates

Years to Mainstream Adoption:

10 + years - Artificial General Intelligence

2 to 5 years - Machine & Deep Learning

Page 8: Strata - Putting AI to work for business - It's a Journey

Top 10 Technology Trend in 2018 – AI Foundation

8

Trend No.1 : AI Foundation– The ability to use AI to enhance decision making,

reinvent business models and ecosystems, and remake the customer experience will drive the payoff for digital initiatives through 2025.

– Narrow AI, consisting of highly scoped machine-learning solutions that target a specific task with algorithms chosen that are optimized for that task, is where the action is today.

– Enterprises should focus on business results enabled by applications that exploit narrow AI technologies and leave general AI to the researchers and science fiction writers"

Page 9: Strata - Putting AI to work for business - It's a Journey

Future of Data Science…Narrow AI – Democratizing Machine Learning

9

1960sDigital

Calculator Spreadsheet SQLMachine Learning

IBM 1980sDesktopIBM

Busi

ness

In

nova

tion

IBM 1990sRDBMS IBM 2010s

Open Source

2018

Machine Learning – Next 2 to 5 Years…

Technology Drivers

Page 10: Strata - Putting AI to work for business - It's a Journey

Understanding AI and Machine learning..

AI is simulation of intelligent behavior in computers

Machine Learning is the science of getting computers to act without being explicitly programmed

Deep Learning is a subset of Machine Learning that leverage neural networks to learn from unlabeled data

Artificial Intelligence

Machine Learning

Data MiningDeep

Learning

Big Data

Page 11: Strata - Putting AI to work for business - It's a Journey

11

Identify Patternsnot recognizable by humans 1

Build Modelsfrom those patterns2

Make PredictionsWith the deployed models3

Machine Learning - Fundamentals

Page 12: Strata - Putting AI to work for business - It's a Journey

Data Science Project - Fundamentals

IBM Cloud / DOC ID / Month XX, 2017 / © 2017 IBM Corporation

Clearly Articulate Use Case

Gather all the Data

Machine Learning

Prepare Data

DigitalApplication

Evaluate

Page 13: Strata - Putting AI to work for business - It's a Journey

13

Artificial Intelligence Transformation…

Articulate Use Case

Break Down Data Silos

Identify AI Technologies, Partner or Acquire tools to fill technology gaps.

Integrate AI in your workflow, between business units

Adopt an open and collaborative culture.

“AI is the new Digital Disruption”2017 Mckinsey

Build a Strategy…

Page 14: Strata - Putting AI to work for business - It's a Journey

Strategies to Consider to drive AI for Business..

Think 2018 / DOC 8111 / March 19, 2018 / © 2018 IBM Corporation

Build a balanced Data Science teamProgrammatic and visual data scienceEnable business domain experts

Connect data, algorithms and applicationsInteractive data access, prep and quality wherever data residesBring tools into one environment

Simplify, accelerate and operationalizeQuick experimentation to deployment to increase business contributionCross-train and increase team velocity

1

2

3

14

Page 15: Strata - Putting AI to work for business - It's a Journey

Math & Stats

ComputerScience

Domain Expertise

Data Science Projects Require multiple Skills

Unicorn

MachineLearning

ResearchEngineering

Scripting, SQL Python, R ScalaData PipelinesBig Data/ Apache Spark Mathematics

Computational

Domain KnowledgeSupply ChainCRMFinancialsNetworking

Data Science Skills

Page 16: Strata - Putting AI to work for business - It's a Journey

How many people with all of these skills exist?

Think 2018 March 2018 / © 2018 IBM Corporation

Source: https://mywebvault.wordpress.com/2017/05/18/modern-data-scientist-skill-set-marketing-distillery/

Page 17: Strata - Putting AI to work for business - It's a Journey

Orchestrated an approach to exploit code-based and visual-based data science from Exploration to Production

Data Preparation

Data ExplorationModel Development

Exploration Production

Implementation

DeploymentModel Management

Fast Iteration

Page 18: Strata - Putting AI to work for business - It's a Journey

Data Science: Time Spent on Activities

~80% of time ~20% of time

Data Collection and Preparation

- Wrong data

- Cleaning data

- Sampling

- Selection Bias

- Feature Engineering

--------------------------------

-

”independent and

identically distributed”

data (idd) is the

statistical gold standard

--------------------------------

Modelling

- Machine Learning

“Systems that can

learn from data”

Evaluation and Deployment

Page 19: Strata - Putting AI to work for business - It's a Journey

Data Science Work: Current Tool Usage

SQL (42%) R

(33%) Python(26%)

Excel(25%)

JavaRubyC++

(17%)

SPSSSAS(9%)

Page 20: Strata - Putting AI to work for business - It's a Journey

Data Science Tools – R & Python Ecosystem..

BUILD

• Monitoring & Alerting

• Model retraining

• KPI Dashboards

• Model Refactoring

• Security

RUNDEPLOY

Experts and Leaders: Sales, Tech Sales, Offering Management, Enablement & Partners

Engaging to WIN: Assets, Services & Marketplace

DATA

XGBoost

jupyter R zeppelinauto prep

model build GUI

object storage databases hadoop

Page 21: Strata - Putting AI to work for business - It's a Journey

The Data Science Project Journey/ Needs

21Think 2018 / March 2018 / © 2018 IBM Corporation

Team/ProjectPhase

Team Tasks and

Challenges

Data Science Leader

Challenges

Getting Started ModelingExperimentation

Developing AppsDeveloping Dashboards

DeploymentMonitoring Support

• Defining projects

• Finding corporate

data

• Connecting to data

sources

• Understanding the

data

• Cleaning data

• Building models

• Measuring

accuracy

• Finding more data

• Building repeatable data

pipelines

• Integration with

engineering

• Machinery management

• QA

• Reuse of old models

• Accuracy monitoring

• Scalability

• Models robustness with

new data

• Integration with

infrastructure

• Reused models in

production

• Hiring, getting skills

• Data security

• LOB understanding

of use cases

• Data security

• Productivity of a very

expensive & rare skill

• Skill inconsistency

• ROI of Data Science

• Data security

• Quality of system

• Productivity of a very

expensive & rare skill

• Knowledge loss due to

high employee turnover

• Auditability &

governance

• Meeting customer

operational expectations

• Productivity of a very

expensive & rare skill

• Knowledge loss due to

high employee turnover

• Auditability & governance

Page 22: Strata - Putting AI to work for business - It's a Journey

Data Science Projects Phases/ Needs

22

Phases: Getting Started ModelingExperimentation

Developing AppsDeveloping Dashboards

DeploymentMonitoring Support

• Defining projects• Finding corporate

data• Connecting to data

sources • Understanding the

data

• Cleaning data• Building models

Measuring accuracy

• Finding more data

• Building repeatable data pipelines

• Integration with engineering

• Machinery management• QA

• Accuracy monitoring• Scalability• Models robustness with

new data• Integration with

infrastructure• (reuse of old models)

• Hiring, getting skills• Data security

(/breaches)

• Data security• Productivity of a very

expensive & rare skill• Skill inconsistency

• Data security• Quality of system• Productivity of a very

expensive & rare skill• Knowledge loss due to

high employee turnover• Auditability & governance

• Meeting customer operational expectations

• Productivity of a very expensive & rare skill

• Knowledge loss due to high employee turnover

• Auditability & governance

Data ScienceLeaders

Data Scientists

Page 23: Strata - Putting AI to work for business - It's a Journey

Accelerate Data Science with a Platform to Scale with Peace of Mind?

IBM Analytics

Page 24: Strata - Putting AI to work for business - It's a Journey

Value of a Data Science Platform

Explore at scale

• Scale out on-demand• No Dev-ops/engineering setup

Reproducibility

• Process of tracking• Reproduce results easily

Secure

• Governed Access• Administration capabilities

Collaborate

• Understand what’s been done• Share and accelerate learning

Publish Efforts

• Models as APIs out of the box• Avoid Engineering re-work

Discovery to Production

• Minimal efforts• Seamless scale• Integration with business process

Open

• Use desired tool of choice• Interoperability across tools

Review Results

• Stakeholder review• Via Dashboards/Static reports

Monitoring

• QA/QC on-demand• Retrain

Page 25: Strata - Putting AI to work for business - It's a Journey

Climb the AI Ladder

2

5

• Leaders will establish a discipline in

data science and accelerate the use

of artificial intelligence

• The “AI Ladder” is an evolutionary

process with various starting points

• IBM Analytics simplifies every step

on the “AI Ladder” by delivering 3

platforms that ensure success

AI

Machine Learning

Analytics

Data

The “AI Ladder”

Page 26: Strata - Putting AI to work for business - It's a Journey

© IBM Corporation 262

6

3 platforms, 1 Information architecture

Data Science& BusinessAnalytics

Hybrid Data Management

Unified Governance & Integration

IBM Analytics Platform help you climb AI ladder

AI

Machine Learning

Analytics

Data

The “AI Ladder”

Page 27: Strata - Putting AI to work for business - It's a Journey

© IBM Corporation 2727

Hybrid

Data Management

Unified

Governance

Data Science &

Visualization

Write Once, Access Anywherewith a common access layer to

promote application independence

Prepare, Publish and Protectyour data to drive insights while

mitigating compliance risks

Descriptive, Predictive, Prescriptive to understand the current, predict the future, and

change the outcome

Machine LearningOn-Premise and CloudOpen Source

Organize Govern Intelligence

Infused with

Seamless between

Powered by

Page 28: Strata - Putting AI to work for business - It's a Journey

© IBM Corporation 28

IBM Analytics Cloud Private – Now with no assembly required

Introducing An engineered platform for doing data science, data

engineering, and app building faster with no assembly required

IBM Cloud Private for DataDATA

ONEInformationArchitecture

Hybrid Data Management

Data Science & Business Analytics

Unified Governance& Integration

Cloud AgilityEasily build data-driven apps, Pre-assembled personalized experiences, Extensible with open APIs

Lightning FastProvision users in minutes, Fast Data ingest speeds,Fast track projects with industry models

AI-readyMachine Learning everywhere, Makes your data ready for AI

App BuildingData EngineeringData Science

DIY : Today, we offer building blocks that can be assembled to do

App Building, Data Engineering, and Data Science.

3 Core Offerings using FlexPoints

Page 29: Strata - Putting AI to work for business - It's a Journey

© IBM Corporation 29

IBM Cloud Private - agility of a unified and trusted platform

IBM Cloud Private

IBM Cloud Private for Data

Common Data & Analytics Services

Data

Cloud-native Data Micro Services

Personalized, Collaborative Team Platform

Works with IBM Watson Studio for AI development, deployment & management

DataScientists

DataEngineers

Business Users, Analysts & CxOs

Data Stewards

App Developers

AI Model Developers

Instant, Pre-assembled Provisioning Admin & Ops Dashboards

Enterprise Data Catalog

ü Data integrationü Data catalogingü Governance & privacy ü Data lifecycle management

ü Customized databasesü Data warehousingü Fast Data Ingest (events)ü Federated query

ü Predictive Modelingü Machine Learningü Mathematical modelsü Dashboards & reporting

Business Partners

Extensible, Open API Platform

IBM Cloud Private for Data

• Amodernenterprisedataplatformbuiltonprivatecloud

• Includestightlyintegrateddata&analyticsserviceswithasimple,collaborativetaskdrivenexperience

• Q2GA– Standard&enterpriseeditions

Page 30: Strata - Putting AI to work for business - It's a Journey

Partner with IBM to drive value from data science investments

30Think 2018 / DOC ID / Month XX, 2018 / © 2018 IBM Corporation

• ExpertisePowered by IBM’s unparalleled data science track

record and know-hows

• EcosystemSupported and extended by rich partner ecosystem

• LicensingFlexible entitlements for business agility and cost-

optimization. Includes subscription and FlexPoint

• End-to-End SolutionComplete, enterprise data science and analytic solution

Business Analytics

IBM CognosIBM Planning

Analytics

Optimized decision

support and prescriptive

analytics

IBM Decision Optimization

Cognitive exploration and content

analytics

IBM Watson Explorer

Visual data science and predictive analytics

IBM SPSS Modeler

Core data science

platform

Watson StudioIBM Data Science

Experience

IBMData Science Platform

+

Page 31: Strata - Putting AI to work for business - It's a Journey

31

Watson StudioIBM Data Science Experience

Community Open Source IBM

• Tutorials

• Data

• Articles and papers

• Fork/ Share projects

• GitHub Integration

• Scala/Python/R/SQL

• Jupyter/ Zepplin Notebooks

• RStudio IDE and Shiny

• Apache Spark

• Your favorite Open libraries

• Visual Aids, Data Flows, Model

Builders

• Dev Ops, Machine Learning

Service, Simple Deployments

• Projects, Version Control

• Security

Watson Studio/ Data Science Experience Core Design

Page 32: Strata - Putting AI to work for business - It's a Journey

Predictive Power

100%

Capacity

Model Builder (CADS)

Build model1

Deploy model2

Refresh model3

Import Sources:§ DSx Notebooks§ DSx Flow UI§ External tools

Auto-generate model from input data, testing various algorithms for best fit (e.g. CADS)

Detect loss of predictive power and refresh model, subject to preferences

Deploy model into production -scale, manage and monitor

Model Automation Model Deployment

Model

Machine Learning Service - Core Elements

Page 33: Strata - Putting AI to work for business - It's a Journey

Web Service

Data Access:• Easily connect to

Behind-the-Firewall and Public Cloud Data

• Catalogued and Governed Controls through Watson Data Platform

Creating Models:• Single UI and API for

creating ML Models on various Runtimes

• Auto-Modelling and HyperparameterOptimization

Web Service:• Real-time,

Streaming, and Batch Deployment

• Continuous Monitoring and Feedback Loop

Intelligent Apps:• Integrate ML

models with apps, websites, etc.

• Continuously Improve and Adapt with Self-Learning

Operationalizing Machine Learning – ML Services

IMS

Page 34: Strata - Putting AI to work for business - It's a Journey

IBM’s Data Science Elite team co-engineer prototypes with clients to succeed and lead in AI via Data Science

What do we offer?ü Up to 3 months’ free onsite engagement

ü Identify use case(s) & Minimal Viable Products via discovery

& design workshops

ü Collaboratively build & evaluate up to 4 Sprints (using IBM’s

Data Science Experience)

ü Mentor & enable client teams hands-on

What do we ask of clients?Dedicated team members to match our headcount on the

engagement

Who are we?A team of IBM Data Science experts, with skills in:

• Descriptive, predictive & prescriptive analytics

• Industry-specific use cases

• Machine learning, deep learning, decision optimization, data

engineering, data journalism

Identify

use case

Break into

4 Sprints

Build

models / pipelines

Validate

approach

Deploy via

APIs

Monitor /

retrain

Visualize

before vs after

dashboard

Validate

with LoB

IBM Data Science EliteDedicated to client

success

Page 35: Strata - Putting AI to work for business - It's a Journey

Thank you

35Think 2018 / DOC ID / Month XX, 2018 / © 2018 IBM Corporation

Carlo AppuglieseProgram Director, Machine LearningIBM Analytics, Data Science Elite [email protected]