Transcript LARC is SMU - Offices and Staff Directory

Updated
29 Nov
2012
LARC is SMU – Carnegie Mellon Partnership
“This extraordinary LARC opportunity takes our existing relationship
with CMU to a new level of research intensity. The SMU-CMU
collaboration gives us global edge in interdisciplinary research that
integrates computation, management, and social sciences.”
- Professor Arnoud De Meyer, SMU President
“The Living Analytics Research Centre builds on CMU’s successful
collaborations with SMU over the years. We are pleased to be partnering
with SMU on such an exciting initiative - one that has great potential for
groundbreaking work in the emerging field of computational social science.”
- Dr Jared L. Cohen, CMU President
2
LARC Project Settings & Partners
3
What is Living Analytics?
Consumer & Social Insights From
Experiment-Driven, Closed-Loop Analytics +
Societal-Scale Human Networks
Framework for Living Analytics
• Observe complex behaviors in natural
consumer and social settings via digital traces
• Progressively real-time
• Progressively societal-scale
• Experiment-driven
• Closed-loop,
and iterative
• Network-centric
5
Closed loop, network experimentation via LARC
1. Collect data (historical, existing, new)
2. Construct network
• Relevant observed links
• Infer links as appropriate
3. Identify questions and predict behavior
• Related to individual behaviour
• Related to group or collective behaviour
• Which method of personalization works best?
4. Design Experiments
• Sample individuals and groups
• Incentive and interaction design
5. Observe via digital traces and interactions
6. Analyze results of network-centric experiments and test
predictions
7. Learn and adapt
8. Iterate around the loop
6
The Living Analytics Adaptive Learning Loop
 The loop begins with the Observe stage that involves observing user interaction and
relationships within a network in real-time and gathering their digital traces.
 The Analyze and Predict stage takes these digital traces, conducts analysis on them,
discovers patterns in them, and uses these patterns for future user behavior and
network trend prediction.
 The Experiment stage involves testing how individual users and networked groups
respond to changes in content, service offerings, interaction experience, pricing and
incentives. The Experiment stage also tests how users respond to different types of
guidance and feedback.
 Finally, the Human Action stage is where users respond within the experiment, and
to various types of feedback, and this generates the data that is picked up on the
next cycle of observation.
7
LiveAnalytics (LARC + LiveLabs) : New Concepts, Methods
and Tools for Consumer & Social Insights that are
LARC
•Experiment-driven
•Closed-loop, and iterative
•Network-centric
•Observing complex behaviors via digital
traces
•Progressively real-time
•Progressively societal-scale
•Combining field realism & complexity
with lab control
Plus LiveLabs
•
Context aware
•
Using real-time context triggers for
automating behavioral interactions
•
Combining usage-adaptive 4G
network management with end-user
behavior
8
LiveAnalytics Vision
(LARC + LiveLabs)
Analytics that combine
realism, complexity and dynamics
of social and consumer behavior
observable in the field
with
experimental control and
causal inference capability of the lab
in a
network-centric world
LiveLabs
9
10
Living Analytics is an Interdisciplinary Fusion of
Computer Science + Social Science
(Computational Social Science)
• Machine Learning & Data Mining
for Data Analytics
• Real-Time Optimization & Adaptive Decision
Support for Decision Analytics
• Social & Management Science for
understanding, predicting and analysing the
behaviour of individuals and networks via
empirical analysis and experimentation
• Enabling computation and software applications
• Enabling privacy and information security
• Enabling protocols and administrative processes
for end-to-end Living Analytics insight
experiments
11
LARC
LARC
Faculty
Faculty
Directors
Directors
and
and
Deputy
Deputy
Directors
Directors
from
from
SMU
SMU and
and
CMU
CMU
12
Funding for LARC
Research grant funds from the Singapore’s Interactive &
Digital Media (R&D) Programme Office.
S$26M
These funds come from Singapore’s National Research
Foundation.
Contributions from SMU, which includes in-kind
contributions as well as financial contributions.
Contributions from Carnegie Mellon, which are comprised
of in-kind contributions.
S$26M
Contributions from external organizations, which can be in
the form of financial or tangible in-kind investments.
$10M
Total (2011 thru 2015)
S$62M
13
Utilization of LARC Funding Over 5 Years
Research grant funds
from the Singapore’s
Interactive & Digital
Media (R&D)
Programme Office
Contributions from
SMU
Contributions from
Carnegie Mellon
Contributions from
external
organizations
Total
• 15 research staffs per year supported at SMU
• 3 full time admin staffs at SMU
• 8 LARC/SIS PhD students spend 10 months at
CMU each year
• Over 5 years, a total of 40 PhD students, each with 10
month training at CMU
S$26M
•
•
•
•
Additional 4 year SIS PhD scholarships (8)
110 months of CMU faculty residency time at SMU
CMU PhD student residency time at SMU
Conference travel and Int’l workshops on LA
S$26M
• LARC facility at SMU
• In-Kind faculty time from SMU
• In-Kind faculty time from CMU
$10M
• Infrastructure
• Field pilots
• Support staff
S$62M
14
“Behavioral Insight”
Experiments:
Methodological Research Work
Experimental methods and tools to
reliably disentangle, quantify and
understand critical effects-e.g. interactions and influences across:
Individual people
Networks of
people
Attributes of
services, content,
experience
Consumer
preferences
Preference
formation &
evolution
Preference
Influencing
How to Design and Execute “Behavioral Insight”
Experiments
• In NETWORKED DATA Context
• Subject to PRIVACY CONSTRAINTS
• At Scale, At Speed, With Smart Use of Network
Resources
Experimental methods, tools and practices for
•Sampling
•Design of interventions
•Execution of interventions, both near-real time
(e.g. on mobile device) and non-real time (e.g.
Web applications)
•Inference, Interpretation, and conclusions
How to design, execute and interpret an
ongoing program of overlapping short-running
& long-running experiments
Human Action:
Individual Responses
Group & Network Responses
For example:
• How will people respond over time to new content, new services,
new interactions, or new bundling or services?
• How will people respond to specific types of incentives in a given
context?
• How will people respond to specific types of price alterations ? How
does this change with context? With time?
• What content will people create ? How does this evolve over time?
• What will people share? How does this evolve over time?
• How will people respond to customized recommendations? To
explanations of consequences ?
• Who do people trust? Who has influence over others in the network?
How does trust and influence evolve over time?
How are these
responses
influenced by
• network
interactions?
• context ?
• learning and
experience
accumulation?
Key R&D Challenges for LARC
1. Analytics, Systems, and Computing Challenges
• Methods and Computation for Observation, Analytics, Prediction, Guidance and
Learning
• In near real-time
• At scale
• Across multiple sources
• Across multiple contexts and extended time periods
• Analytics at individual level of granularity while preserving privacy
• Economically so that firms can eventually afford to make use of these powerful
capabilities
2. Social & Management Science Challenges
• Understanding behavioural choices, preferences and intent of individuals,
in the context of networks of individuals, depending on context
• Designing & delivering guidance and incentives
• Evaluating how participants respond to guidance & incentives
• The theory and practice of experimental design in evolving, network-centric
settings as observations cannot be assumed to be independent
17
Hard Challenges to Realising Living Analytics, con’t
3.
Administrative Challenges
• Gaining access to the data of participating companies
• Legal agreements to work with companies (confidentiality, publishing, IP issues,
exclusivity, no warranties, indemnities, etc)
• Gaining permissions to work with participating communities
• Institutional Review Board (IRB) clearances
• Privacy and security issues for working with confidential data and human subjects
• Training industry counterparts (project side, legal side, business side) on how to
work with the university, on accelerating capability development vs procuring
“product”
• Training LARC faculty, staff and students to understand the criticality of these
legal agreements and giving them tools and processes to abide by requirements
o Practical and future oriented solutions to all of these challenges are part of LARC’s work.
o SMU and CMU are giving these issues the highest levels of senior management support.
o These types of administrative capabilities are important strategic competencies
that enable LARC to work with private and public sector organisations.
18
Assuring the Security, Privacy and Confidentiality
of our Partner Information
19
Post-graduate PhD
students who enter
our programme
within the next two
years wills be elgible
to participate in the
10 month training
residency at CMU
through LARC
20
Engaging with LARC
21
Engaging with LARC
Data Set Partner • Commitment to joint experimentation with LARC
• Data access for LARC (SMU & CMU)
• Major investment required for specific project
initiative
• Need additional manpower at SMU and
CMU since all manpower is committed to
ongoing projects
• Ability to sponsor and co-supervise
Research
LARC-related Student Projects with SMU and
Affiliate Member
•
•
•
•
CMU students
Annual private briefing
Periodic affiliates workshop
Preferential access to students and LARC team
Invitations to LARC seminars and events
22
Living Analytics and LARC
23
LARC: experimentation and learning in a
digitally connected, network centric world
24
Acknowledgments
LARC Data Set Partners
• BuzzCity
• Citibank Singapore
• Starhub
• ResortWorlds Sentosa
• Sentosa Leisure Group
•LiveLabs Urban Lifestyle
Innovation Platform
LARC Research Affiliates
25
Contacting LARC
For any enquires or more information about
LARC, please contact us at 6808 5227
or email to [email protected]
Visit us at http://www.larc.smu.edu.sg/
Like us at http://www.facebook.com/larc.cmu.smu
Follow us on https://twitter.com/larc_cmu_smu
26