R.G. (Randy) Goebel is currently professor of Computing Science in the Department of Computing Science at the University of Alberta, Associate Vice President (Research) and Associate Vice President (Academic), and founding principle investigator in the Alberta Machine Intelligence Institute (AMII). He received the B.Sc. (Computer Science), M.Sc. (Computing Science), and Ph.D. (Computer Science) from the Universities of Regina, Alberta, and British Columbia, respectively. Professor Goebel's theoretical work on abduction, hypothetical reasoning and belief revision is internationally well know, and his recent research is focused on the formalization of visualization and explainable artificial intelligence (XAI). He has worked on optimization, algorithm complexity, systems biology, and natural language processing, including applications in legal reasoning and medical informatics. Randy has previously held faculty appointments at the University of Waterloo, University of Tokyo, Multimedia University (Kuala Lumpur), Hokkaido University (Sapporo), visiting researcher engagements at National Institute of Informatics (Tokyo), DFKI (Germany), and NICTA (now Data61, Australia); is actively involved in collaborative research projects in Canada, Japan, China, and Germany.

1. Randy Goebel, Professor of Computing Science at the University of Alberta
From Data to Economic Value

2. Like many potentially impactful technologies, the positive economic impact of Artificial Intelligence
(AI) is challenged by both strategic and practical misunderstandings. Line of sight from science to
value is required to help achieve positive impact.
Speaker: Randy Goebel,
Alberta Machine Intelligence Institute, Edmonton, Canada
Volkswagen Data Lab, Munich, Germany
From Data to Economic Value

3. Federal AI Strategy Development
https://medium.com/politics-ai/an-overview-of-national-ai-strategies-2a70ec6edfd

4. World Ranking
(CSRankings.org)

5. The “Big
Data”
value
chain
decision
support
Data
Infor-
mation
Knowledge Action
IoT
sensor
networks
…
Security-
access federation
…
integration
domain
knowledge
machine learning
modeling
…
resource-
bounded choice
action sets
…
Continuous improvement by creation of learning systems

6. Data Science
Workflow
Define Objectives
• Confirm
access to
appropriate
data
• Define clear
domain specific
KPIs
Data Capture &
Curation
• Acquire data from
original sources
(e.g. sensors or
legacy systems)
• Ensure possibility
for continuous
capture, correction,
and adjustment
Data Analysis
• Perform data
analysis to explore
data value
potential
E.g,. time series
cycles, principle
component
analysis, data
labelling/annotatio
n requirements
Data Model Generation
• experiment with
alternative model
generation
techniques
E.g., logistic
regression, support
vector machines
(SVM), deep
learning,
Reinforcement
Learning,
condition-action
rules, etc.
• Create
Interactive
Dashboard for
Domain Experts
• Design
Evaluation
Experiments to
Measure
performance
against KPIs
Data Model Interaction
Interaction, Feedback, and Adjustment as required
Key success requirements:
1. access to data
2. data domain expertise, including how to measure
potential value for related business processes, and
3. data scientists committed to work with the data
domain experts to achieve improvements in domain
expert defined key performance indicators.

7. Where does
data science
apply?
• Personal and public (health) data
• Repertoire of changing intervention
choices
• Feedback loop for intervention outcomes.
Precision Health
Precision Law
Precision Education
Precision Democracy

8. The data
value
chain in
medicine
Outcome
assessment
Data
Infor-
mation
Knowledge Action
IoT
sensor
networks
…
Security-
access federation
…
integration
domain
knowledge
machine learning
modeling
…
Health
economics
Clinical practise
guidelines
…
Continuous improvement by creation of learning systems

9. Data and “Personalized” Medicine
Towards personalized medicine …
U
U
U
…

10. Data and Precision Medicine
Towards precision medicine …
. . .
. . .

11. Image-based
glioma biopsy
• Input:
• - ?? labeled cases
• Method:
• compute S-transform frequency
spectrum for each pixel, the average over
all directions to get 1D spectra for each
tumor;
• build multi-frequency classifier based
on these labeled cases
• in these measurement results, BLUE 30
co-deleted 1p/19q, GREEN 24 intact
1p/19q
• Results:
• - 95-96% accurate

12. Image-based
glioma biopsy
• Input:
• - 55 labeled cases
• Method:
• compute S-transform frequency spectrum
for each pixel, the average over all directions
to get 1D spectra for each tumor;
• build multi-frequency classifier based on
these labeled cases
• in these measurement results, BLUE 30 co-
deleted 1p/19q, GREEN 24 intact 1p/19q
• Results:
• - 95-96% accurate

13. Summary
Keeping abreast of AI Science to quickly develop
and apply AI technologies
• Strengthen connections between science and
industry
• Keep attention on social impact of AI and
emerging societal values
Economic Value of AI – like anything, what’s the
ROI?
• Based on identifying and extracting models
which support business models and economic
KPIs
Impact obtains from connecting good research
and good business expertise
• Access to (good) data
• Access/collaboration with domain experts
• Collaboration/commitment of data scientists