COM S 490 Examples Examples of COM S 490 Projects/Studies  PEPPER, Frans Effendi '05 (Spring 2005)
Orthogonal Procrustes and Rotation of Subspaces, Wei Guo '05 (Fall 2004) Publish-Subscribe System, Muthiah M Muthaia Chettiar '06
(Fall 2004) Home Health Horoscopes, Jinen Kamdar '04 (Spring 2004)
Robocup 2004, Ivan Han Hor Siu '06 (Spring 2004) Comparison of Agglomerative and Network Flow-based Clustering Methods, Omar Nayeem '04 (Spring 2004) Mostly-Unsupervised Statistical Segmentation of Simplified Chinese Sequences,
Alex Cheng '05 (Spring 2003) User Satisfaction in the
NSDL, Ross Housewright '05 (Spring 2003) Using First-Order Logic to Reason About Policies,
Ilya O. Ryzhov '05 (Spring 2003) An Analysis to Determine the Relative
Commutativity of Operations on Nested Data Structures,
Brian Hackett '04 (Spring 2003) RoboCup
2003, Yuval Shavit '03 (Spring 2003) Computer
Game Design (Part 1),Chihiro Fukami '04, David Lin '06, Dustin
Shultz '04, Eric Holmberg-Weidler'04, Gregory Evan Poucher '05, Joseph
Egbulefu '04, Jonathan Lyons '05, Lorraine Pace '05, Louis Lu '03, Sunil
Annapareddy '05, Loraine Sydney de Lapeyrouse '04, Wei-Han Ho '05 (Spring
2003) Computer
Game Design (Part 2),Gene Pang '04, AJ Fink '04, Brandon Irizarry
'03, Hwan-Ting Lee '03, Jacqueline Bodine '04, Laura Johnson '06, Linda Nguyen
'05, Xiu Xu '05, Ryan Pijai '05, Simon Leet '05, Zennard
Sun '04, Eugene Lee '04, Alexander Fierro '05(Spring 2003) A
Space-Optimized Java Virtual Machine for Mobile Ad Hoc Nodes, Matthew
Wachs '04 (Spring 2003) Multi Dimensional Scaling on
3D models, Saumya Vardhan Kanoria '03(Fall 2002) Investigation
and Implementation of Advanced Rendering Techniques, Eugene Lee '04 (Fall 2002) Creating Tools to Animate 3D
Effects, Michael Ferguson'03 (Fall 2002) Accent
Restoration Decision List vs Hidden Markov Models,Ivn Canales Saldaa '05 (Fall 2002) Using multiple homologous
sequences during sequence to structure alignment of proteins,Craig Lowe '04 (Fall 2002) Analysis of
Braess's Paradox in network routing,Henry Lin '02 (Fall 2002) Investigation of
Computer-Game Design: Part 1,Jacqueline Bodine '04, Eli Brown '04, AJ Fink '04, Simon Leet '03, Antal
Novak '03, Nikita Pavlov '04, Zennard Sun '04, Xiu Xu '05 (Fall 2002) Investigation of
Computer-Game Design: Part 2,Dan Andreescu '04, Gregory Fodor '03, Brandon Irizarry '03, Alexander Krol
'04, Navin Kumar '03, Eugene Lee '04, Hwan-Ting Lee '03, Andrew Gould '05 (Fall 2002) Natural
Language Processing: A Natural Langauge Based Review,Philip Zigoris '03 (Fall 2002) Economics from an
Algorithmic Perspective, Daniel Golovin '03 (Fall 2002) Analysis of Ad Hoc
Mobile Wireless Networks,Linda
Nguyen '05 (Fall 2002) Development
of a Short Course in C#, Radha Narayan '05 & Howard Kwong '03 (Spring 2002) Feasibility
Study of a 3D Graphics Engine, Dan Andreescu '04 and Melissa Whitcomb '05 (Spring
2002) Interval
Analysis, Optimization, and Ecology in Matlab, Justin Tung '03 (Fall
2001) Parser generation in the Java age, Aleksey Kliger
'02 (Fall 1999) Radiosity techniques for realtime simulations on low-end hardware,
Jason Rohrer '03 (Fall 1999) RoboCup'99, Scott Aaronson '01
(Fall 1998) Client-side
Database Applications, Victor Cheng '99 (Fall 1998) Distributed On-line Ticket Sales,
Ernie Gremillion '99 (Fall 1998) Bali++ Compiler, Ryan Kennedy '99 (Fall 1998) Computational Biology, David Liben-Nowell '99 (Fall 1998) Neural Networks/Reinforcement Learning, Alex Rakhlin
'00 (Fall
1998)
| top | Abstracts and Notes for COM S 490 Projects/Studies PEPPER, Frans Effendi '05 (Spring 2005)Instructor: Johannes Gehrke
Abstract: PEPPER is a Peer-to-Peer Data Management System. As part of the
PEPPER project, we are exploring a much richer query semantics for
peer-to-peer systems. We envision a future where users will use
their local servers to offer services described by semantically-
rich XML documents. Users can then query this "P2P service
directory" as if all the services were registered in one huge
centralized database. Towards this goal, we are developing novel
peer-to-peer indexing and query processing techniques. In this
project, I am involved in the overall design and
implementation of the PEPPER system. Orthogonal Procrustes and Rotation of Subspaces, Wei Guo (Fall 2004) Charles Van Loan Given m by n matrices A and B, the Orthogonal Procrustes problem looks into finding an n by n orthogonal matrix that minimizes the Frobenius norm of A-B*U. I will be focusing on rotation of subspaces and in particular, various algorithms that numerically compute an orthogonal rotational matrix U for a given matrix B, which sends B to B*U that is closest to matrix A in terms of minimum Frobenius norm of A-B*U.
Publish-Subscribe Systems , Muthiah M Muthaia Chettiar (Fall 2004) Johannes Gehrke RPlus Tree is a dta structure for multi-dimensional objects. I will be implementing an improvement over this tree, RPlus Tree With Holes, specifically tailored to be fast and efficient for Publish Subscribe systems, and benchmark its performance.
Home Health Horoscopes, Jinen Kamdar (Spring 2004) Instructor: Phoebe Sengers Abstract: The Home Health Monitor is a design for a system that provides feedback about a home's emotional temperment. Light, temperature, door sensors and other indicators can be used to study people's activities inside the different areas of their home. "Sensor readings and trends are mapped in state space to a collection of sentences drawn from a large number of published horoscopes to automatically generate a bespoke horoscope for the particular household on a given day" (source: Gaver, Beaver and Benford,"Ambiguity as a Resource for Design"). In other words, we want to provide people with a systematic but menaingful reflection of the emotional state of their home. Aside from the technical aspects of this project (using an XML database to generate a corpus of horoscope info and machine learning to determine the emotional tone of a sentence, and then selecting the appropriate sentences), the HCI involves studying horoscopes and their impact on people.
Robocup 2004, Ivan Hor Siu Han (Spring 2004) Instructor: Bart Selman Abstract: Robocup is a multi-disciplinary project which aims to build fully autonomous robots capable of engaging other teams of robots in a game of soccer. I will be assisting the COM S team in developing strategy in general and offensive maneuvers in particular.
Comparison of Agglomerative and Network Flow-based Clustering Methods, Omar Nayeem (Spring 2004) Instructor: John Hopcroft Abstract: In Spring 2003, I determined the running time of an agglomerative clustering algorithm to help determine its scalability to data sets larger than the Citeseer database, for which it was primarily designed. I found that the algorithm was not as scalable as previously thought, and I worked to optimize it. In Fall 2003, I aided in the design and implementation of an algorithm using a clustering approach based on network flows, which I work to optimize in Spring 2004.
Mostly-Unsupervised Statistical Segmentation of Simplified Chinese Sequences,
Alex Cheng '05 (Spring 2003) Instructor: Lillian Lee
Abstract: This project focuses on the implementation of the TANGO algorithm based on a paper by Rie Kubota Ando and Lillian Lee. The purpose of this project is to examine the effectiveness of this algorithm to Simplified Chinese Sequences.
| return to List | User Satisfaction in the NSDL,
Ross Housewright '05 (Spring 2003) Instructor: Bill Arms
Abstract: Working to implement features in the NSDL portal to allow tracking of user behavior, and to analyze this behavior to determine the quality of searching provided by the portal. Ultimate goal of the project is to work on determining a good way to balance
fulltext searching with metadata searching over metadata of variable quality.
| return to List | Using First-Order Logic to Reason About Policies,
Ilya O. Ryzhov '05 (Spring 2003) Instructor: Carl Lagoze
Abstract: We are building the front-end of an application that takes policies from users, converts
them to first-order logic, and then proves in the logic that certain people are permitted or forbidden to do
specific actions. Although rough sketches exist for the GUI, my partner and I will be responsible for hammering
out the details and both designing and implementing the underlying data structures. We will, of course, work
with the back-end team to make sure that the data is translatable and we may also do some pre-analysis to
reduce the about of information that must be translated to answer a given question such as `may Alice download
the report?'
| return to List | An Analysis to Determine the Relative
Commutativity of Operations on Nested Data Structures,
Brian Hackett '04 (Spring 2003) Instructor: Radu
Rugina
Abstract: Two operations are relatively commutative if the order in which they are run
does not affect the result of running a specific third operation afterwards. The initial goal of
this project is to develop an analysis which can automatically determine the relative
commutativity of a limited set of insertion/extraction operations, although arbitrary heap operations should
ultimately be considered.
| return to List | RoboCup 2003, Yuval Shavit '03
(Spring 2003) Instructor: David
Schwartz
Abstract: The purpose of the RoboCup project is to develop a system of fully autonomous
robots to play soccer. As a member of the COM S team, I will be developing and implementing
strategies for use in the 2003 Artificial Intelligence system. I will also be part of the move to make the system fully autonomous so
that even setting up the players is computer controlled.
| return to List | Computer Game
Design (Part 1) Chihiro Fukami '04, David Lin '06, Dustin Shultz '04, Eric
Holmberg-Weidler'04, Gregory Evan Poucher '05, Joseph Egbulefu '04, Jonathan
Lyons '05, Lorraine Pace '05, Louis Lu '03, Sunil Annapareddy '05, Loraine
Sydney de Lapeyrouse '04, Wei-Han Ho '05 (Spring
2003) Instructor: David Schwartz, Rama C. Hoetzlein
Abstract: This project is an exploration of the
processes of computer game design. In addition to technical topics in
artificial intelligence, computer graphics, physics and music, we will explore
the history of computer games, the game development process, and violence in
media and video games. Students will be expected to
explore original game ideas and learn to work in groups
while learning new techniques. The last seven weeks we
will develop final game projects in collaboration with
students in music and art.
| return to List | Computer Game
Design (Part 2) Gene Pang '04, AJ Fink '04, Brandon Irizarry '03, Hwan-Ting
Lee '03, Jacqueline Bodine '04, Laura Johnson '06, Linda Nguyen '05, Xiu Xu '05,
Ryan Pijai '05, Simon Leet '05, Zennard Sun '04, Eugene Lee
'04, Andrew Gould '05, Alexander Fierro '05(Spring
2003) Instructor: David Schwartz, Rama C. Hoetzlein
Abstract: This project
course is an exploration of the processes of computer game design and related
technical concepts in animation, computer graphics, artificial intelligence,
game networking, physics and music. Students will work in groups to develop
projects of their own choosing. Weekly progress in group projects will be
motivated by the students, and will ideally involve a mixture of technical and
aesthetic content. Groups will be expected to create a week-by-week plan for
each team member in the group. Final projects will be presented and evaluated
by other groups in the class and possibly by the Part 1 students.
| return to List | A
Space-Optimized Java Virtual Machine for Mobile Ad Hoc Nodes, Matthew Wachs
'04 (Spring 2003) Instructor: Gun
Sirer
Abstract: This project comprises porting a
preexisting, third-party Java Virtual Machine to the
StrongARM architecture, with an emphasis on reducing the
environment's memory footprint. Optimization techniques will
include conventional practices, such as examination of data structure
alternatives, spooling, and memory allocation deferment; however,
additional efficiency gains will be realized through use of the
MagnetOS distributed operating system to amortize storage of requisite
data structures across a network of mobile ad hoc nodes.
| return to List | Multi Dimensional Scaling on
3D models, Saumya
Vardhan Kanoria '03 (Fall 2002) Instructor: Steve
Marschner
Abstract: The goal
of this project is to try and determine the different dimensions along which
people judge the quality of 3D models. An experiment will be carried out where
various people will be asked to identify/differentiate between sets of
decimated 3D models rendered using different materials in different
environments. Multi Dimensional scaling will be done on this data to
try to find the dimensions along which people judge the quality of these
models. The results of this experiment will be useful in trying to create more
perceptually intuitive light reflection models.
| return to List | Investigation
and Implementation of Advanced Rendering Techniques, Eugene Lee '04 (Fall 2002) Instructor: Kavita
Bala
Abstract: The goal of this independent study is
to gain a better understanding of global illumination algorithms by
implementing advanced rendering techniques such as photon mapping. Over the
semester, a software renderer with Photon Mapping, capable of generating
global illumination solutions, will be implemented. In addition, techniques
such as irradiance caching, projection maps or two-pass Photon Mapping with
caustics may be investigated and incorporated into the rendering software.
| return to List | Creating Tools to Animate 3D
Effects, Michael Ferguson'03 (Fall 2002) Instructor: Kavita
Bala
Abstract: I am creating a software package for
creating 3D animations. Users will be able to transform a model differently at
key-frames, and the software will interpolate to create the rest of the movie.
Transformations will include opening up a model and slicing up a model. This
software will be tailored to work with large models which are created by
scanning real-life objects, and so it will expect the models to have color
data stored in each vertex. In order to render previews at interactive speeds,
this software will use level of detail algorithms to simplify the models.
| return to List | Accent
Restoration Decision List vs Hidden Markov Models,
Ivn Canales Saldaa '05 (Fall 2002) Instructor: Lillian
Lee
Abstract: This project with focus in
implementing and comparing two different approaches (Log-Likelihood Decision
List and Hidden Markov Model) to the same problem (Accent Restoration in
Spanish corpora) based on the papers written by David Yarowsky and Ya'akov
Gal. The purpose is to determine what characteristics in the training
corpora modify the accuracy and recall of both algorithms.
| return to List | Using multiple homologous
sequences during sequence to structure alignment of proteins,Craig Lowe '04 (Fall 2002) Instructor: Ron
Elber
Abstract: This study focuses on evaluating the
performance increase when multiple sequences are used to find homologs in a
list of known structures. Instead of doing a sequence to structure alignment
with only the input sequence against the list of known structures, the program
will first find sequences of high sequence homology to the input sequence and
align those to the list of known structures as well. I will also be looking at
metrics which may be used to quantitatively evaluate performance increases in
the program.
| return to List | Analysis of Braess's
Paradox in network routing,Henry
Lin '02 (Fall 2002) Instructor: Eva
Tardos
Abstract: This project focuses on doing
additional research on Braess's Paradox. The paradox is
the idea that in a selfish network, deleting an edge can
actually improve overall performance. Essentially,
Braess's Paradox arises because "bad" links can be added so that
selfish users begin clogging a link that was originally
fast, and thus slow down performance for everyone. The project this semester will involve writing a program to calculate
selfish and globally optimal routing schemes and using
it to further characterize Braess's Paradox (e.g. how
often it occurs, how bad, etc.).
| return to List | Investigation
of Computer-Game Design: Part 1,Jacqueline Bodine '04, Eli Brown '05, AJ Fink '04, Simon Leet '03, Antal Novak '03,
Nikita Pavlov '04, Zennard Sun '04, Xiu Xu '05 (Fall 2002) Instructor: David
I. Schwartz
Abstract: This project is an exploration of the
processes of Computer Game design. General topics to be explored include the
history of computer games, character development, the game development
process, and violence in media and video games. Students will analyze
the game design process, develop game ideas, and construct projects in a
variety of game genre, focusing on both abstract ideas and specific technical
challenges. Topics in computer science include Finite State Machines (AI),
Computer Graphics, and Physical Simulation.
| return to List | Investigation
of Computer-Game Design: Part 2,
Hwan-Ting Lee '03, Dan Andreescu '04, Eugene Lee '04, Brandon Irizarry
'03, Alexander Krol '04, Gregory Fodor '03 (Fall 2002) Instructor: David
I. Schwartz
Abstract: This project is an interdisciplinary
exploration of the process of Computer Game design. Students in computer
science and the arts will work in creative teams to develop projects in
parallel. General topics to be explored include character development, the
history of computer games, and violence in media and video games. Game
projects will involve specific topics in both computer science and the arts.
Topics in the arts include Character Development, Motion Studies, Cel
Animation, and Music. Topics in computer science include Finite State Machines
(AI), Computer Graphics, and Physical Simulation. Students will be expected to
work in their area of expertise to complete collaborative assignments and
projects.
| return to List | Natural
Language Processing A Natural Language Based Review,
Philip Zigoris'03 (Fall 2002) Instructor: Lillian
Lee
Abstract: This independent study will be a
discussion based review of current work in the field of Natural Language
Processing. There will be a focus on statistical and machine learning methods.
Discussion is not necessarily geared towards any final project but instead on
giving critical thought to current research.
| return to List | Economics
from an Algorithmic Perspective,Daniel
Golovin '03 (Fall 2002) Instructor: Jon
Kleinberg
Abstract: Broadly, I will be investigating
algorithmic aspects of various economic models of
markets. Of particular interest are the properties of equilibria,
such as their existence and complexity, and properties of various
market models, such as their efficiency according to various metrics.
| return to List | Analysis
of Ad Hoc Mobile Wireless Networks, Linda
Nguyen '05 (Fall 2002) Instructor: Gun
Sirer
Abstract: This project investigates the
efficiency of various routing algorithms in the context of an ad hoc mobile
wireless network. Routing investigations will be performed on a testbed
consisting of a combination of about 100 handhelds and laptops. In addition to
routing, topology/energy control, network clustering, ad-hoc internetworking,
and more, will be studied. An analysis program will be built to aid in
analysis.
| return to List | Development of a Short Course in
C#,Radha Narayan '05 and Howard Kwong
'03 (Spring 2002) Instructor: David Schwartz
Abstract: This project will involve the
investigation of C# and .Net technology's effectiveness as a teaching
language. By learning the language, the researchers will develop
teaching materials for a 4-to-6 week course. As part of the study, the
researchers will illustrate any advantages and disadvantages in the language
and teaching thereof.
| return to List | Feasibility Analysis of a 3D Graphics
Engine, Alexander Krol, Chester Hsieh, Christopher
Hynes, Dan Andreescu, Matthew Herndon, Melissa
Whitcomb, Navin Kumar, Nikita Pavlov, Patrick Dowell, and Dylan Tong (Spring
2002) Instructor: David Schwartz
Abstract: In a small team, we will research
3D graphics engine design: the mathematics involved, the choices of design
environments, the available libraries, existing techniques, and anything else
that may give us an edge. As part of a continuing project, we will first
lay the groundwork for the implementation of a 3D engine as a game or research
tool.
| return to List | Interval Analysis, Optimization,
and Ecology in Matlab, Justin Tung '03 (Fall
2001) Instructor: David Schwartz
Abstract: Interval analysis is an uncertainty method for deterministic mathematical models.
In this project, we will apply IA to an ecological model concerning
foraging. We want to optimize a condition given by the model using fixed
point (non-interval) and interval algorithms in Matlab. Variations and
constraints on model parameters will be explored using IA. Also,
numerical analysis of interval and fixed point root approximations will be
done with applications to the model.
| return to List | Parser generation in the Java
age, Aleksey Kliger '02 (Fall 1999) Instructor: Andrew Myers Abstract: This project will
focus on implementing a parser generator for the Java programming language. The
goal of the parser generator is to facilitate a Java programmer unfamiliar with
writing a parser to create a parser using syntax similar to that of Java itself.
In addition, in case of ambiguity in the grammar, the parser generator will
signal errors to the programmer in the form of sample ambiguously parsable
sentences.
| return to List | Radiosity
techniques for realtime simulations on low-end hardware, Jason Rohrer '03 (Fall 1999) Instructor: Thomas Yan
Abstract:I am developing a multi-platform (ANSI standard C++) non-realtime
radiosity rendering engine. I am also developing a multi-platform (Mac and Win32) realtime
3D simulation engine that will depend on the radiosity engine's calculations for realistic
lighting and shading. I am using the 3DFX Glide graphics acceleration API to implement the
realtime portion of the package.These techniques should produce realtime simulations with
very realistic lighting and shading. | return to List |
RoboCup'99, Scott Aaronson '01 (Fall 1998) Instructor: Bart Selman
Abstract: I'm head of Strategy & Vision for Team Italy,
one of Cornell's two teams
building autonomous mobile robots for the RoboCup'99 robotic soccer
competition to be held in Stockholm this August. I'm mainly working on
algorithms for multi-agent collaboration, motion planning, prediction,
and other computational tasks needed to successfully defend and score goals.
| return to List | Client-side Database Applications, Victor Cheng '99 (Fall 1998) Instructor: Praveen Seshadri Abstract: I am working with PhD student Tobias Mayr in developing
a Java User-Defined Function application for the Predator database system. This will allow
Java methods to be incorporated into database queries on Predator. I will also be involved
in designing demos of applications throughout their phases of completion.
| return to List | Distributed On-line Ticket
Sales, Ernest Gremillion '99 (Fall 1998) Instructor: Ken Birman Abstract: This independent study is an extension of a project done
for COM S 514. It will be a client/server application that sells tickets for multiple events
at different venues over the web.
| return to List | Bali++ Compiler, Ryan Kennedy (Fall 1998) Instructor: Keshav Pingali Abstract: Bali is a procedural-oriented subset of the
Java programming language. We are working to redesign the Bali language, producing a new
object-oriented subset of Java, known as Bali++. We will then redesign and implement a
Bali++ parser and code generator. Finally, we will be make the necessary additions to
Bali++'s associated assembly language, SAM, to enable the object-oriented features of
Bali++. | return to List |
Computational Biology, David
Liben-Nowell (Fall 1998) Instructor: Jon Kleinberg Abstract:We are exploring various models for
determining the distance between two genomes (a continuation of last semester's work). | return to List |
Neural Networks/Reinforcement
Learning, Alex Rakhlin(Fall
1998) Instructor: Bart Selman Abstract:
Neural networks can be used to learn values of
game states. I looked at how neural networks can learn a simple 3x3 tic-tac-toe game.
First, the problem of loading examplars was considered and different setups were used to
load the minimax values of all states in the shortest amount of time (the problem is
NP-complete in general). Data was obtained on how learning time varies with parameters
such as the number of nodes, learning rates, momentum, etc. Second, the generalization
problem was considered. The resulting networks showed a good generalization ability. The
data obtained suggests that the best approach for loading examplars is the "separate
agents" approach. This modular approach is currently believed to be most promising
for developing learning agents. | return to List |
| top | Last updated September 14, 2004. For more information email us at ugrad@cs.cornell.edu .
|
