The Center's research is structured as an overlapping sequence
of PROBEs (PROBlem-oriented Explorations). Each PROBE focuses
on a specific problem domain, and lasts for approximately one
year. Its purpose is to find algorithmic solutions and appropriate
mathematical formulations for that problem domain, and thus involves
both domain experts as well algorithm designers.
The PROBE topics are selected to have some common themes, so
that ideas are shared among the PROBEs. Each PROBE provides funding
for faculty, students, and possibly a postdoctoral fellow, as
well as travel funding for the workshops. To maximize pay-off,
PROBEs seek broadly-applicable solutions to domain problems, and
not specialized solutions for specific scenarios nor detailed
solutions for the purpose of commercialization (i.e., research,
not product development).
PROBEs
Flexible Network Design
Spectral Graph Theory and Applications
Algorithms for SNP and Haplotype Analysis
Algorithms in Economics
Computer-Human Authentication with Applications to AI
Dynamic Algorithms and Applications
Graph Cuts for Vision and Data Analysis
Integrated Logistics
Meshing, Theory and Applications
Privacy In D.A.T.A.
Scheduling Your Network Connections
Utilizing the Power of Human Cycles
Web Structure and Algorithms
PROBE selection
The PROBE topics are selected by the Scientific Advisory Committee.
We encourage people to submit proposals. A proposal should be
from 2-3 pages and contain the following:
- An overview of the domain
- The current state-of-the-art in algorithms applied to the
domain.
- The set of algorithms/problems that the PROBE will explore
- A list of PROBE organizers (from 2-4).
The PROBEs will be selected based on the following three criteria:
- Potential impact
- The idea of a PROBE is not to explore something known to work,
but rather to explore something which might work and can have
significant impact.
- Cross-disciplinary participation. The suggested participants
must include members from other institutions and must include
both domain experts and algorithm designers. Furthermore, it
must also contain some people who can help bridge the gap. Ties
to other PROBEs. There should be connections between the PROBE
and other existing or potential future PROBEs.
- PROBE budget The budget can vary from PROBE to PROBE and we
are always looking for some matching support from other sources.
Assuming no external funding, a PROBE will cover about 3.5 faculty
summer salaries, 2.5 students, .5 postdocs, $10K travel, and
$15K "workshop" costs. This, however, can be balanced in various
ways.
PROBE Schedule
In general The PROBEs run in a synchronized schedule, from June-to-June.
Each PROBE is punctuated by three workshop-style meetings, each
lasting from one to three days. In the first meeting (in June),
the participants gather for tutorials on the topic, and to initiate
a discussion between the domain experts and algorithm designers.
This is expected to jump start a productive summer of collaborative
research. The second meeting (in October) allows the same participants
to discuss progress to date, including detailed proposals of algorithms
and techniques to apply within the domain, and new theoretical
problems to evaluate. These algorithms, techniques and problems
are explored and evaluated in further detail between the second
and third meetings. The third meeting (in June) presents results
of the research. While the first and third meetings are open to
anyone,the second only involves the main participants.
PROBE results
Each PROBE produces several concrete results, including published
papers, running code, pages for our web repository, data sets,
and case studies for the education component. In addition to these
concrete results, there are many other benefits of each PROBE.
These can include new interesting theoretical problems, new ways
to characterize data, new ways of formulating and viewing problems
within the domains, and the transfer of knowledge and techniques
across domains. Also, as the term PROBE and its expansion imply,
it is our expectation that most of the PROBEs will lead to longer-term
studies of the respective areas. These longer term studies will
not be directly funded as part of the Center, but infrastructure
within the Center will help find funding for these follow-on studies,
especially from industry.
Each PROBE will assign a student scribe and a student code manager.
The Scribe will be responsible for documenting the problems and
algorithms that were explored. This will be used to help the organizers
write the case study document. The Code Manager will be responsible
for collecting data sets and algorithms developed as part of the
PROBE.
PROBEs will be expected to produce the following concrete results:
(1) tutorials covering the domain, and covering relevant algorithms
in the domain, (2) a case study document, (3) code and data, (4)
a set of web pages on the topic, (5) publishable papers, and (6)
a course on the topic.