  |
 |
 |
 |
 |
 |
 |
 |
 |
Past Lunch
Bunches
2008 | 2007 | 2006 | 2005
Tuesday,
January 22, 2008
Jason Marden and Adam Wierman
74 Jorgensen
A
Game Theoretic Formulation of the Sensor Allocation Problem
Interdisciplinary research at the junction of information sciences, economics
and game theory offers new solutions to several problems. This talk looks at
applying game theory to solve a problem in cooperative control in distributed
systems: the optimal locations of sensors in a sensor network.
This talk is introductory. The problem domain and basic concepts will be described. Detailed proofs will be given in other seminars and are also found in papers.
This talk presents a view of cooperative control using the language of learning in games. Specifically, we look at the cooperative control problem of dynamic sensor allocation. We formulate the sensor allocation problem as a noncooperative game where the decision makers are the sensors. In this setting, each sensor is assigned a local objective or utility function and is given the ability to autonomously alter it's position and sensing activity in real time. The distributed nature of the decision making allows for robustness to communication failures, sensor failures, and environmental changes. In this talk we will discuss several methods for designing the sensors' local utility functions. We measure the efficacy of a particular utility design in two ways:
(i) Does
a Nash equilibrium exist
(ii) How efficient is a Nash equilibrium when compared to the optimal
allocation?
Tuesday,
January 29, 2008
Joel W. Burdick, Caltech
74 Jorgensen
Engineering
Interfaces to Damaged Nervous Systems
This
talk will introduce some of the challenges involved in trying to develop
technology that can partially restore some functionality to people with
severe neural deficiencies. The first part of the talk (which is based
on joint work with Prof. Richard Andersen and Prof. Y.C. Tai of Caltech)
will focus on neural prostheses. A neuroprosthetic is a brain-machine
interface that can potentially enable a paralyzed human, via the use
of surgically implanted electrode arrays, and associated computer decoding
algorithms, to control external electromechanical devices. The problem
of decoding the prosthetic user's intent from the recorded neural sisgnals
is a problem in inference, or estimation. A few of the estimation techniques
that have been used to solve this problem will be reviewed. Progress
on making adaptive electrodes that autonomously optimize the quality
of the brain-machine interface will also be reviewed.
The second half of the talk (which is based on joint work with Prof. Reggie Edgerton of UCLA and Prof. Y.C. Tai of Caltech) will focus on the problems involved in partially restoring locomotion after a severe spinal cord injury. While there are an enormous number complex issues surrounding a spinal injury, this talk will focus on the use of drug therapy, automated machines for physical therapy, and a new class of epidural spinal cord stimulators to help with the locomotion rehabilitation process. Each of these therapeutic components also have associated computational and algorithmic problems. For example, we are currently trying (in collaboration with Prof. Yaser Abu-Mostafa of Caltech) to formulate the problem of selecting the correct electrode stimulation protocols as a problem in kernel learning.
Tuesday,
February 5, 2008
Paul Upchurch,
Caltech & JPL
74 Jorgensen
Game
Engines for Interactive Visualizations of Space Exploration
The number of people playing online games continues to increase each
year. Game engine technology is generating innovations in hardware and
software. This project investigates the use of game engines at JPL for
interactive visualizations of explorations of the solar system.
This talk describes GoView a scriptable game engine which is tailored for rendering spacecraft trajectories in the solar system. Rendering trajectories across the solar system poses challenges that are not encountered in popular online games such as Doom. In particular, the game engine must be able to handle spatial scenes larger than six orders of magnitude. The talk describes the problems encountered and how they are solved in GoView.
Interactive visualizations of explorations of the solar system are being developed at JPL for public outreach, mission planning and mission operations.
Tuesday,
February 12, 2008
Erik Winfree,
Caltech
74 Jorgensen
Learning
to Program Chemistry
Biological
organisms are extremely sophisticated self-organized chemical systems.
Their complexity dwarfs anything produced by modern chemical industries.
The difference can be ascribed in large part to biological systems
being information-processing machines: DNA, RNA, and proteins carry
molecularly-encoded messages that direct all of life's processes. Biology
is programmable chemistry. Can we learn how to program chemistry? How
will this expand the capabilities of synthetic chemistry? Might non-biological
chemistries also be programmable? What kinds of models of computation
are needed for understanding how to program chemistry and the limits
to doing so? I cannot give conclusive answers to these questions, but
I will present examples of what we have learned while creating self-assembling
structures, molecular machines, and logic circuits built out of DNA,
RNA, and the occasional enzyme. I will argue that the key to unleashing
the revolutionary potential of programmable molecular systems lies
in understanding the design space and managing their complexity as
information processing systems -- issues that are fundamental to computer
science.
Tuesday,
February 26, 2008
Paul
De Martini, Southern California Edison
74 Jorgensen
Partnering
with our Customers for a Smarter, Cleaner Energy Futu
SmartConnect,
a new effort by Southern California Edison, will enable the utility
to manage increasing demand for electricity from customers and
provide environmental benefits by deploying a smart meter system
which could reduce peak demand by as much as 1,000 megawatts -
the output of a large power plant - as customers reduce some peak
electricity usage and shift some peak usage to off-peak periods
of the day when power costs less. Additional savings include lower
labor costs due to the use of wireless data transfer from meters
to the utility rather than manual meter reading. Paul De Martini,
Director of SmartConnect, will describe the program, its development
plan, and the technological challenges it faces.
Tuesday,
March 4, 2008
Hsuan-Tien
Lin, Caltech
74
Jorgensen
From Ordinal Ranking to Binary Classification
Ordinal
ranking is an important concept in modeling our preferences. We
rank hotels by stars to represent their quality; we give feed-backs
to products on Amazon using a scale from one to five; we say that
the weather is hot, warm, cool, or cold without referring to the
actual temperature. The wide applications of ranking range from
social science to behavioral science to information retrieval.
For yet another example, in 2006, Netflix (an on-line DVD rental
company) announced a million-dollar-prize challenge for building
a better automatic personalized movie ranking system, and the prize
is heating up the competition in machine learning and related areas.
Many machine
learning approaches are designed in recent years to understand ordinal
ranking better, but the design process can be time-consuming. Our work
presents a novel alternative -- a reduction framework that systematically
transforms ordinal ranking to simpler yes/no questions, i.e., binary
classification.
Then, well-studied binary classification approaches can be effortlessly
casted as new ordinal ranking ones. Furthermore, the reduction framework
reveals a strong theoretical connection between ordinal ranking and binary
classification, and allows us to easily extend well-known theoretical
results for binary classification to new ones for ordinal ranking. In
this talk, I will discuss the intuition and the construction of the reduction
framework, as well as its theoretical and algorithmic merits, using the
Netflix challenge as an example.
Tuesday,
March 11, 2008
Jeanne
Holm, JPL
74
Jorgensen
Virtual
Worlds and Space Exploration
Populations
of virtual worlds, such as Second Life, have grown rapidly. This talk
discusses using virtual worlds for bringing lots of people into the
NASA mission, letting them participate in the day to day work and successes
of the U.S. space agency. NASA has established several islands in Second
Life. NASA CoLab and Explorer Island are the two main public entrance
points. These areas are geared at working with any person who is interested
in learning more about NASA or, better yet, participating in a NASA
mission for exploration in this virtual world. While NASA CoLab focuses
on a place to host meetings and talks, Explorer Island (created by
the Jet Propulsion Laboratory) is meant to be an immersive environment
for interacting with spacecraft, being at a live launch of the Space
Shuttle or a mission to Mars, talking to NASA scientists and engineers,
sharing your ideas for space exploration in international workshops,
and walking on the surface of another world. Come in world and look
for Jet Burns (Charles White) or Devery Barrymore (Jeanne Holm) and
we'll give you a tour!
Tuesday,
March 18, 2008
Ulrich
Pinkall, Berlin University of Technology
12:00
pm, 74 Jorgensen
What
Game Engines Can Do For Mathematical Visualization
Surfaces with constant (mean or Gaussian) curvature are a classical topic
in Differential Geometry. The shape of such surfaces is usually quite
complex and poses a real challenge for Computer Graphics and interactive
visualization. We will demonstrate that an optimal environment for doing
mathematical experiments with surfaces is to put them in a virtual landscape
where one can walk on the surfaces and interact with them in a way modeled
after first person computer games.
Tuesday,
April 1, 2008
Azita Emami, Assistant Professor of Electrical Engineering, Caltech
12:00 pm, 74 Jorgensen
High-speed
Interconnects in Modern VLSI Systems
The implementation
of high-performance computing systems strongly relies on the feasibility
of high-bandwidth data communication between integrated circuit components
(IC's). Moreover, future multi-core processors will need fast and robust
intra-chip data transfer between the cores and memory units. Current
trends of digital systems indicate that the amount of processing of
each component or unit is expected to continue to increase exponentially
at least for the next 10 years. In order to scale the communication
bandwidth with the same trend, both the number of IO's and the data-rate
per IO link need to increase. A number of limitations such as the bandwidth
of the wires, area and power consumption per IO, interferences, and
characteristics of the highly scaled devices make the design of high-speed
IO's very difficult.
This talk will focus on low-power system and circuit solutions for parallel chip-to-chip interconnections and intra-chip networks. As the data-rates over the conventional wires increase, the complexity of required equalization and coding schemes increase rapidly. The design challenges of wireline data communication and the possibility of using optics for interconnection at short distances will be discussed. We will focus on a number of novel low-power solutions for both electrical and optical signaling, and the scaling properties these solutions for the future systems.
Tuesday,
April 8, 2008
Shankar Kalyanaraman, CS Graduate Student, Caltech
12:00 pm, 74 Jorgensen
It
Is (NP-)Hard To Rationalize Marriages
Given a set of observed economic choices, can one infer preferences and/or
utility functions for the players that are consistent with the data?
Questions of this type are called rationalization or revealed preference
problems in the economic literature, and are the subject of a rich body
of work.
From the computer science perspective, it is natural to study the complexity of rationalization in various scenarios. We consider a class of rationalization problems in which the economic data is expressed by a collection of matchings, and the question is whether there exist preference orderings for the nodes under which all the matchings are stable.
We show that the rationalization problem for one-one matchings is NP-complete. We propose two natural notions of approximation, and show that the problem is hard to approximate to within a constant factor, under both. On the positive side, we describe a simple algorithm that achieves a 3/4-approximation ratio for one of these approximation notions. We also prove similar results for a version of many-one matching.
Tuesday,
April 15, 2008
Chris Umans, Caltech, Computer Science
12:00 pm,
74 Jorgensen
Randomness
and Pseudorandomness: The Computational Perspective
Computational
complexity views randomness as a resource to be conserved, much like
running time or storage space, while "pseudo-randomness" is
defined in terms of fooling a computationally-bounded observer. These
unique persepectives lead to a number of widely applicable and powerful
techniques, and form the basis for one of the most active areas of
complexity theory.
In this talk I'll give a tour of some of these techniques -- randomness-efficient sampling, explicit constructions, and pseudo-random generators -- with example applications in game theory, coding theory, data structures, and complexity.
Tuesday,
April 22, 2008
Tom Heaton, Caltech Prof. of Engineering Seismology
12:00 pm, 74
Jorgensen
Designing
the Next Generation of Seismographic Network
I will briefly describe the design and capabilities of the Southern California
Seismic Network that is a cooperative project of Caltech’s Seismological
Laboratory and the U.S. Geological Survey. This network consists of several
hundred digital stations that continuously telemeter data to a central
site for processing and archival. The current technology allows rapid
access to widely distributed ground motion information over an extremely
wide range of amplitudes (200 dB) and frequencies (30 to 0.001 Hz). However,
the relatively small number of stations means that is not possible to
observe unaliased seismic wavefields in either the ground or buildings.
I will discuss the possibility of increasing the spatial density of seismographic
stations in southern California by several orders of magnitude. This
involves development of a distributed seismic network where instruments
are maintained by affiliated agencies and individuals
Tuesday,
April 29, 2008
Krishna Palem,
Rice University
12:00 pm, 74 Jorgensen
What To Do
About The End of Moore's Law (Probably)?
Many claim
that the laws of physics dictating the exponentially improving benefits of Moore’s
Law will end in the next 10 to 20 years. The argument for the end of Moore’s
Law is based, in part, on an analysis that switching devices cannot function
deterministically as feature sizes get reduced to molecular levels. Moore’s
Law could, however, continue provided systems with probabilistic switches could
process information usefully. My research suggests that this is indeed possible
in contexts where the "quality" of the results of the computation is
perceptually determined by our senses—audio and video information being
significant examples. To demonstrate this principle, I will show how CMOS-based
devices, circuits and computing architectures whose correctness is characterized
probabilistically can be used effectively. I will show that significant (a multiplicative
factor of 2 or more) energy and performance gains can be achieved, while trading
a perceptually tolerable level of error—through probabilistic adders and
multipliers, applied in the context of finite impulse response filters and FFT
engines processing video and audio data in digital signal processing. Quantifying
the human tolerance for error, we expect, will be ultimately based on neurobiological
models. Conceptually, our thesis recommending tolerating error in the switching
devices in return for savings in cost, is analogous to a parametric extension
of Simon's notion if satisficing—we will conclude the talk by dwelling
on this analog and its implications to probabilistic design of ultra large-scale
integrated (ULSI) circuits in the future.
Tuesday,
May 6, 2008
John Doyle,
Caltech
12:00 pm, 74 Jorgensen
Rules
of Engagement: The Architecture of Robust, Evolvable Networks
Biological
systems are robust and evolvable in the face of even large changes in
environment and system components, yet can simultaneously be extremely
fragile to small perturbations. Such universally robust yet fragile (RYF)
complexity is found wherever we look. The amazing evolution of microbes
into humans (robustness of lineages on long timescales) is punctuated
by mass extinctions (extreme fragility). Diabetes, obesity, cancer, and
autoimmune diseases are side-effects of biological control and compensatory
mechanisms so robust as to normally go unnoticed. RYF complexity is not
confined to biology. The complexity of technology is exploding around
us, but in ways that remain largely hidden. Modern institutions and technologies
facilitate robustness and accelerate evolution, but enable catastrophes
on a scale unimaginable without them (from network and market crashes
to war, epidemics, and global warming). Understanding RYF means understanding
architecture — the most universal, high-level, persistent elements
of organization — and protocols. Protocols define how diverse modules
interact, and architecture defines how sets of protocols are organized.
Insights into the architectural and organizational principles of networked systems can be drawn from three converging research themes. 1) With molecular biology’s description of components and growing attention to systems biology, the organizational principles of biological networks are becoming increasingly apparent. Biologists are articulating richly detailed explanations of biological complexity, robustness, and evolvability that point to universal principles. 2) Advanced technology’s complexity is now approaching biology’s. While the components differ, there is striking convergence at the network level of architecture and the role of layering, protocols, and feedback control in structuring complex multiscale modularity. New theories of the Internet and related networking technologies have led to test and deployment of new protocols for high performance networking. 3) A new mathematical framework for the study of complex networks suggests that this apparent network-level evolutionary convergence within/between biology/technology is not accidental, but follows necessarily from the universal system requirements to be efficient, adaptive, evolvable, and robust to perturbations in their environment and component parts.
Tuesday,
May 20, 2008
Yury
Lifshits, Caltech
12:00 pm, 74 Jorgensen
The
Architecture of the Web
Initially
the Web was defined as a system of interlinked hypertext documents
accessed via the Internet. Its structure is regulated by World
Wide Web Consortium via standards like URL, HTML, CSS and RSS.
The Web of 2008 is a system of data, web applications and people.
Necessity of syncronization raised the next generation of standards:
OpenID, OAuth, Social Graph API, Microformats and various widget
platforms (F8, Open Social, Firefox, iPhone, iGoogle, Wordpress,
Drupal, Salesforce.com).
We start with systematic overview of Web architecture. Then we discuss research agenda in the field: data flow (publishing, synchronization and exchange), web operating systems and identity management. Finally we present our ongoing project "New Approaches to Online Marketing".
Tuesday,
May 27, 2008
Jeremy
Ma, Caltech
12:00 pm, 74 Jorgensen
An
Inside Look at the Multi-Agent Robotics Lab
Mobile agent communication and control is a widely studied area in robotics,
with application in various fields of science. The Multi-Agent Robotics
Lab in the Mechanical Engineering Department consists of several mobile,
robotic agents each equipped with sensing, on-board computing, and wireless
communication. In this talk, a brief introduction to the testbed, its
capabilities and applications will be presented.
Tuesday,
May 27, 2008
12:00
pm, 74 Jorgensen
Alumni
from industry will talk about their experiences and give advice.
Career
Planning and Ice Cream
Students
have asked speakers to address the questions listed below.
Questions for alumni from industry:
- Does
it help to get a PhD if your goal is to work in industry? (Do you
plan to get an advanced degree later? For example MBA, MS or PhD?
How difficult is it to return back to school after a few years?)
- How helpful
is research experience when looking for industrial jobs? (because
students often have trouble deciding between internships and SURF)
- Could
you tell us more about educational support from most companies for
Master's degrees?
- What
is the process of switching groups or positions?
- Coming
out of your undergraduate studies, what plans did you have for your
career? Did it turn out the way you planned? How have your career
changed since then
- What
are the tradeoffs between working for a very small startup versus
working for an established company?
- Do you
have advice for students planning to start their own companies?
- Does it help to take "business" courses or economics courses if someone wants to enter Wall Street (finance)?
Questions for graduating seniors:
- How
did you decide which companies to interview with? How should we decide?
- Do you
think doing an internship with a company is more helpful than a SURF
in getting an industrial job?
- What
sorts of questions are asked at interviews? How should we prepare
for the job interviews and applications?
- Are companies
looking for specific courses? Specific experiences? Programming projects?
Programming languages?
- Does
being from Caltech help? Hurt? Make no difference?
- How did
you decide whether to go to grad school or industry?
- Do you
plan to attend graduate school later? For an MBA, MS or PhD?
- Could you tell us more about educational support from most companies for Master's degrees (if you know about this from the companies that you have interviewed/worked with)?
| top |
