New Directions in IT Society Research - the IEEE Information Theory ...

photohomoeopathΤεχνίτη Νοημοσύνη και Ρομποτική

24 Νοε 2013 (πριν από 3 χρόνια και 7 μήνες)

74 εμφανίσεις

New Directions in IT
Society Research

Jeff Andrews, UT Austin

Andrea
Montanari
, Stanford

Michelle
Effros
, Caltech

Olgica

Milenkovic
, UIUC

Alex
Dimakis
, UT Austin

Lara
Dolecek
, UCLA

Muriel
Medard
, MIT

Sriram

Vishwanath
, UT Austin


ISIT 2013

Background


IEEE has asked its societies to come up with a document
detailing “Future Directions”


This would presumably go into an IEEE level report


Gerhard asked me to lead this effort


We could also use the outcome for our own purposes,
inc.

IT
Society newsletter, seeding funding agency ideas, etc. (ideas
welcome)


Seems a useful exercise, albeit a challenging one


I
f all this was clear and obvious, we’d have a lot more open
faculty positions and NSF funding for information theory


Related to our Outreach endeavors


I formed the committee, aiming for diversity of research areas,
blend of youth and senior people, top minds in variety of topics

Plan


Emphasize the generality and universality of IT as a serious
scientific discipline


Tout its past triumphs and established intersections with and
contributions to other fields


Provide
c
oncise conjecture (total report < 10 pages) on areas for
potential growth, new synergies, and articulate exciting open areas


Avoid buzzwords, stick to fundamentals


This requires more vision than any small group of people can
provide


W
e approach this with humility, and greatly appreciate inputs from the
BoG

and beyond


Ideally, we can advance a conversation that will be helpful

An Inspiration and Framework


Develop a 2013 update of
such a figure


Articulate existing
intersections (with a bit
more detail)


Conjecture on future
intersections

Figure 1 of Cover and Thomas

(1991)

Current Outline/Ideas


Communications


Mostly reviewing past triumphs


Networking


Fundamental properties of large (general)
networks and graphs


Nano
-
circuits, distributed systems


Control theory


Signal Processing


Compressed sensing


L
ossy

c
ompression,
inc.

for huge data sets


Implementation of IT
-
inspired ideas


Human information acquisition


Physics


Statistical physics, entropy


Quantum information theory



Statistics and Learning Theory


Includes application to enormous data sets


E.g. High
-
dimensional statistics, PCA


Computer Science


Seen as a major area of blurring with us (list
decoding, security, etc.)


Computation as a constraint?


Genetics and Molecular Biology


DNA detection, processing, computing


Virology


Neuroscience


Encoding, storage, processing of
information in neural networks


Economics and Finance


(See graph theory above)


Universal investment theory

Discussion/Questions for
BoG


Are key areas missing? Can any be combined?


To cite or not?


Is this a worthy exercise in
your opinion?

Definition of Information
Theory (back up slide)

Definition.

Information theory is a mathematical
science that studies the ultimate limits of, and optimal
methods and algorithms for:

1. The representation of information;

2. The communication of information;

3. The processing and utilization of information.