MACHINE INTELLIGENCE 4.0
For Globalization 4.0 based on Industry 4.0
By Prasad Chitta
CONTEXT – WHAT IS “4.0” ALL ABOUT?
• Industry 4.0 - https://en.wikipedia.org/wiki/Industry_4.0
• Steam Engine, Electricity, Internet and 4.0
• Globalization 4.0 - https://www.weforum.org/globalization4
• Collaboration, Patriotism, Fair Economy, Environment, Work, social impact
• Business 4.0 - https://sites.tcs.com/bts/digital-transformation-to-
business-4-0-pov/
• Digital transformation for getting businesses ready for 4.0
DATA IS THE NEW OIL – THAT NEEDS
PIPELINES
• From Acquisition to
purging
• transaction
management (OLTP)
• Analysis of data (OLAP)
• Traditional before 4.0
Sensing (Manual,
IoT)
Acquiring,
Validating,
Munging
Storing /
Publishing /
Streaming
Update / Enrich /
Tag
Operational
Reporting,
Dashboards (MI)
Transforming,
Aggregating, De-
normalizing
OLAP reporting
visualizing &
Analytics
Machine
Learning, Deep
Learning
Archiving,
Purging
SYSTEMS DEALING WITH THESE PIPELINES
PIPELINES OF “LEARNING” AND
“INTELLIGENCE”
• Structured data comes from traditional
systems of records
• It can be originated in a data center,
private or public cloud
• Non structured data comes from mail,
document / content management
systems and systems of engagement like
social media
• Semi structured data comes from
machine generated sources and IoT
sources
• Data that comes is processed on a
batch, micro-batch or individual
messaging level
• Data pipelines gradually enrich the data
for better insights
TYPES OF TRAINING (OR) LEARNING(?)
https://towardsdatascience.com/types-of-machine-learning-algorithms-you-should-know-953a08248861
PLATFORMS, TOOLS, LANGUAGES
WHAT IS MACHINE INTELLIGENCE 4.0?
Input
•Program
Machine
1.0
Output
Input
•Output
• Rewards
•Policies
Machine
AutoML
+ Model /
data
Repository
4.0
Auto
Tuned
Model
Input
•Configuration
Machine
With ERP
Software
2.0
Output
Input
•Output
(Labels)
Machine
+ Data
Scientists
3.0
Learning
Model
MACHINE LEARNING – SKILL AREAS
Statistics
Story
telling
Artisti
c skills
Business
Oper
ations
Archit
ectur
al
Soluti
on
devel
opm
ent Tools Enabler
Store
streamProcess
Business KPI
Optimization
Excellence
Data
Scientists
ML Engineers
ML Experts
The challenge
Analyze/
Visualize
Describe
Predict Prescribe
Generate
TWO MODELS OF LEARNING “MACHINE
LEARNING”
Statistics /
Probability /
Linear Algebra
Python,
Tensorflow,
Pytorch
Business domain
/ Problem
domain
Business Problem
and solution
understanding
Auto ML and
Search for APIs
integration into
solutions on cloud
/ platforms
DO IT YOURSELF
A recent competition on kaggle… (A kernel walk through!)
https://www.kaggle.com/c/two-sigma-financial-news
THANK YOU
https://www.linkedin.com/in/
prasadchitta

Machine intelligence 4.0 public

  • 1.
    MACHINE INTELLIGENCE 4.0 ForGlobalization 4.0 based on Industry 4.0 By Prasad Chitta
  • 2.
    CONTEXT – WHATIS “4.0” ALL ABOUT? • Industry 4.0 - https://en.wikipedia.org/wiki/Industry_4.0 • Steam Engine, Electricity, Internet and 4.0 • Globalization 4.0 - https://www.weforum.org/globalization4 • Collaboration, Patriotism, Fair Economy, Environment, Work, social impact • Business 4.0 - https://sites.tcs.com/bts/digital-transformation-to- business-4-0-pov/ • Digital transformation for getting businesses ready for 4.0
  • 3.
    DATA IS THENEW OIL – THAT NEEDS PIPELINES • From Acquisition to purging • transaction management (OLTP) • Analysis of data (OLAP) • Traditional before 4.0 Sensing (Manual, IoT) Acquiring, Validating, Munging Storing / Publishing / Streaming Update / Enrich / Tag Operational Reporting, Dashboards (MI) Transforming, Aggregating, De- normalizing OLAP reporting visualizing & Analytics Machine Learning, Deep Learning Archiving, Purging
  • 4.
    SYSTEMS DEALING WITHTHESE PIPELINES
  • 5.
    PIPELINES OF “LEARNING”AND “INTELLIGENCE” • Structured data comes from traditional systems of records • It can be originated in a data center, private or public cloud • Non structured data comes from mail, document / content management systems and systems of engagement like social media • Semi structured data comes from machine generated sources and IoT sources • Data that comes is processed on a batch, micro-batch or individual messaging level • Data pipelines gradually enrich the data for better insights
  • 6.
    TYPES OF TRAINING(OR) LEARNING(?) https://towardsdatascience.com/types-of-machine-learning-algorithms-you-should-know-953a08248861
  • 7.
  • 8.
    WHAT IS MACHINEINTELLIGENCE 4.0? Input •Program Machine 1.0 Output Input •Output • Rewards •Policies Machine AutoML + Model / data Repository 4.0 Auto Tuned Model Input •Configuration Machine With ERP Software 2.0 Output Input •Output (Labels) Machine + Data Scientists 3.0 Learning Model
  • 9.
    MACHINE LEARNING –SKILL AREAS Statistics Story telling Artisti c skills Business Oper ations Archit ectur al Soluti on devel opm ent Tools Enabler Store streamProcess Business KPI Optimization Excellence Data Scientists ML Engineers ML Experts The challenge Analyze/ Visualize Describe Predict Prescribe Generate
  • 10.
    TWO MODELS OFLEARNING “MACHINE LEARNING” Statistics / Probability / Linear Algebra Python, Tensorflow, Pytorch Business domain / Problem domain Business Problem and solution understanding Auto ML and Search for APIs integration into solutions on cloud / platforms
  • 11.
    DO IT YOURSELF Arecent competition on kaggle… (A kernel walk through!) https://www.kaggle.com/c/two-sigma-financial-news
  • 12.