CCS355 Neural Network & Deep Learning Unit II Notes with Question bank .pdf
Global AI Conference Presentation - Machine Learning for SMB
1. Machine Learning (ML) and automation for the rest of
us (Small/Medium Established Businesses)
Rich Jolly
VP, Data Science
Hawes Group
January 19, 2018
@rich_jolly
Global Artificial Intelligence Conference
Santa Clara, CA
2. Key Messages
• Machine learning and AI can
help, and are within reach of
the small/medium established
business
• 80/20 – Target effort
• Use distributed resources to
gain traction (and save costs)
• Embrace SW best practices
• Evolve into analytics excellence
Luddites
3. • Consumer financial engagement company
– Third-Party collections (primary)
– First-party collections (expanding)
– Technology partner (software platform and services)
• Over 80 years in business, approximately 250 FTEs
• Verticals: healthcare, government, financial services
• Growing national footprint
4. Small Business in US
• 28.8 million Small
Businesses
• 56.8 million Small
Business Employees
• 99.7% of US
Businesses
• 48.0% of US
Employees Source: SBA Office of Advocacy (sba.gov)
5.
6. Recent advances transform Data Science
Compute availability
Rich, open source software
Powerful, open source
machine learning libraries
The capabilities that a few years ago only the big players could
achieve, are now within the reach of most companies!
8. Build a virtuous cycle
Successful
Analytics
Projects
More
organizational
confidence
More
resources
Building analytic capabilities is an evolutionary process
9. Human Resources
VP, Data Science
Direct
Resources
Analysts
Behavior Science
Partners Contractors
Gig
Economy
FIVERR, GURU, etc.
Ad Hoc resources
Vetted data
scientists
Analysts
SW resources
Able to deal with
more ambiguity
Projects must be
tightly encapsulated
10. Pareto Rules!
• A great deal of machine
learning benefit can be
realized with reasonable
effort!
– Don’t forget regression
• Rank opportunities
– Pick low hanging fruit
• Monitor progress
– And results
11. Example: Prioritizing Accounts
Python
Logistic Regression Python
Statsmodels
Natural Language
Processing of agent
comments
NoSQL Database
Bayesian updating
N3090316:10:21 X CI FROM 1234567890 … VRFYD DOB … VRFYD ADDY, RN, NO POE … ADVISED X OF OB
… X SD ACCT NOT VALID … XSFR TO HD
SQL Database MySQLMongoDB
Python,
SciKit Learn,
NLTK
Classify comments:
- Collector or system note?
- Connect? Right party?
- Payment agreement?
12. Example: Supporting Call center agents
• Recording, transcription
and analysis of call center
• Turn Key capabilities:
– Monitor compliance
– Agent performance
• Add on:
– Behavioral Science analysis
Tools Used: VoizTrail
13. Automation
Descriptive Diagnostic Predictive Prescriptive
Gartner Difficulty Scale
RPA – Robotic
Process
Automation
Machine Learning
based AI
Hawes use case: File
transfer and manipulation
Hawes use case: NLP
evaluation of call transcripts
for Behavioral analysis
Tools used: RoboDX Tools used: SciKit Learn, NLTK
14. Embrace SW Best Practices
• Version Control
• COTS vs. DIY
• Agile
• Data cleansing
Image source: Wikipedia
15. Just do it!
Source: BCG
https://www.bcg.com/publications/2017/infographic-what-is-holding-back-artificial-intelligence.aspx