Machine Learning Demystified Vishwa Kolla Head of Advanced Analytics John Hancock Insurance
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Machine Learning Demystified
Vishwa Kolla Head of Advanced Analytics
John Hancock Insurance
TAI
User
Group
, 2017
Technology (CS)
Analytics (Math, Stats)
Business (MBA)
Advanced
Analytics CoE,
Maturity Model
Customer Analytics
(entire value chain)
Machine Learning
Scoring Engine
Optimization
Simulations
Forecasting & Time
Series
• 16+ Years
• John Hancock Insurance
• Deloitte Consulting (Industries – Insurance,
Retail, Financial, Technology, Telecom,
Healthcare, Data)
• IBM
• Sun Microsystems
Expertise
Experience
Vishwa Kolla Head of Advanced Analytics
John Hancock Insurance, Boston
MBA Carnegie Mellon University
MS University of Denver
BS BITS Pilani, India
What? Why? How?
AI, ML, DL, Data Science, Advanced Analytics…
Are all the same
Are very different
Not sure
AI, Data Science, ML, DL …
Are all the same
Are very different
Not sure
Are related,
but different
In terms of evolution, what is the right order?
A. ML -> DL -> AI
B. AI -> DL -> ML
C. AI -> ML -> DL
D. DL -> ML -> AI
In terms of evolution, what is the right order?
A. ML -> DL -> AI
B. AI -> DL -> ML
C. AI -> ML -> DL
D. DL -> ML -> AI
TAI
User
Group
, 2017
How are all of these related?
Source: h2o.ai
Computer Science (CS)
The study of automating
algorithmic processes that scale
Artificial Intelligence (AI)
An ideal intelligent machine is a
flexible rational agent that
perceives its environment and
takes actions that maximize its
chance of success at an
arbitrary goal
Machine Learning (ML)
The study and construction of
algorithms that can learn from
and make predictions on data
Deep Learning (DL)
A branch of machine learning
based on a set of algorithms that
model high-level abstractions in
data using multiple processing
layers
TAI
User
Group
, 2017
Evolution
https://blogs.nvidia.com/blog/2016/07/29/whats-difference-artificial-intelligence-machine-learning-deep-learning-ai/
TAI
User
Group
, 2017
Data Science
Analytics (Math)
Technology (CS)
Business (BBA/MBA)
TAI
User
Group
, 2017
Advanced Analytics
Business Data Math Implement
Internal External
Merge Profile
Segment Explore
Campaign
Execution
Nudge
Videos
Ops
Integration
Apps Applications BI
Strategy Insights Recos
Monitor
Geo-Spatial
What? Why? How?
TAI
User
Group
, 2017
Gartner Hype Cycle 2015
TAI
User
Group
, 2017
Gartner Hype Cycle 2016
TAI
User
Group
, 2017
The promise is real
Value from Internal Data is … well, HUGE
~$820 B
Value of
Customer Data +
Algorithms
$1.2 T Market Cap
(11/30/2016)
$120 B Debt
(11/30/2016)
$178 B Brand Value
(05/2016)
Source: http://www.forbes.com/powerful-brands/list/2/#tab:rank
Embracing Data helps with Top & Bottom lines
2001 – 2013 CAGR Revenue (Firm | Industry)
Source: 2001 – 2013 Revenue figures from Capital IQ
3%
3%
3%
1%
5%
7%
7%
8%
10%
12%
What? Why? How?
TAI
User
Group
, 2017
Machine Learning (ML) gives
computers (machine)
ability to learn (learning)
without being explicitly programmed (learning)
Arthur Samuel, 1959
TAI
User
Group
, 2017
ML techniques
Machine Learning
Supervised Learning
Classification
SVM
Discriminant Analysis
Naïve Bayes
Nearest Neighbor
Regression
Linear, GLM
Trees
(RF, GBM)
Ensemble
Neural Networks
Un-supervised Learning
Clustering
K-Means /
K-Medioids
Hierarchical
Neural Networks
TAI
User
Group
, 2017
Un-supervised Learning Techniques
Source: Machine Learning eBook by Matlab
K-Means K-Medoids
Hierarchical SOMs
TAI
User
Group
, 2017
Un-supervised learning Applications
TAI
User
Group
, 2017
Supervised Learning
Source: Machine Learning eBook by Matlab
Prospect Acquire Nurture
TAI
User
Group
, 2017
ML Use Cases in Life Insurance
Prospecting Nurture Acquisition
Market
Segments
Customer
Segments
Likely To [*]
Media
Mix
Channel
Strategy
Survey
Analytics
Cross-Sell
OCR
Engines
Mortality
Risk
Morbidity
Risk
Stratified
Models
Loss Ratio
APS
Summary
Smoker
Likelihood
Churn
Models
Audience
Propen-
sities
Claim
Severity
Customer
Journey
Litigation
Likelihood
Customer
Engagem
ent
Fraud
Detection
>>
Text
Analytics
Optimi-
zation
Simu-
lations
Recruiting
Analytics
IoT
Analytics
TV
Audience
Analytics
Anomaly
Detection
>>
TAI
User
Group
, 2017
New Use Cases in Insurance – Age, BMI
TAI
User
Group
, 2017
New Use Cases in Insurance – Age, BMI
TAI
User
Group
, 2017
DL – the next frontier
Vision Context
Transcription Translation
29
TAI
User
Group
, 2017
Sources and Acknowledgements
1. Gartner Hype Cycle. http://www.gartner.com/newsroom/id/3412017
2. Deep Learning at Google. https://www.wired.com/2016/02/ai-is-changing-the-technology-behind-google-searches/
3. WSJ. Economic Value of AI. https://blogs.wsj.com/cio/2017/04/28/lower-prediction-costs-the-simple-economic-value-of-artificial-intelligence/
4. John McCarthy. Father of AI. http://www.asiapacific-mathnews.com/04/0403/0015_0020.pdf
Related Documents
