USRP PROJECT TITLES
Name of supervisor: Lua KT
The project title: Information Retrieval Based On Word Senses
IC288 or IC388 or both: both
Brief description of the project (optional):
In this project, we will explore one approach to potentially achievewide
coverage, high accuracy word sense disambiguation (WSD). Clustering
algorithms will be used to find groups of related words from a large corpus
of English texts. This is followed by automated WSD of the related words in
a group. These two procedures can be done off-line to generate a useful
list of related words with disambiguated senses. Such a list can then be
used to disambiguate words in new, previously unseen texts. Such a list can
then be used to disambiguate words in new, previously unseen texts.
Name of supervisor: Lua KT
The project title: Information Theory and Meaning - K T Lua
IC288 or IC388 or both: both
Brief description of the project (optional):
We have found that the information content (entropy) has some effect on the
word construct and its meaning. A word lose information when it is combined
from a number of characters. It is a process of losing entropy for the
binding of the word. We will continue the research to define a scope for
the word sense using its information content.
Name of supervisor: Lua KT
The project title: A Study Semantics of Chinese Words - K T Lua
IC288 or IC388 or both: both
Brief description of the project (optional):
In this project, we attempt to classify meanings in a n-layer tree. The
so-called semantic distances are the distance of the meaning on the tree.
Using this model, we have obtained some important result for Chinese
language. We are able to predict the meaning of Chinese compound at an
accuracy of above 70%. We have also explained why there are only about
5,000 Chinese characters in use. Many more interesting results will be
derived in the near future. We are also working in two directions: (i) to
incorporate results from psychological research about Chinese characters
into the model and (ii) to enhance the scheme of semantic representation to
a net. This work will be continued in the direction of re-clsutering of
meaning of Chinese/English words. A general comparison between word meaning
clustering of Chinese and English words will also be carried out.
Name of supervisor: Lua KT
The project title: Application of Neural Net In the Predicition of Meaning
of Compound Words
IC288 or IC388 or both: both
Brief description of the project (optional):
A three layer back propagation neural net is set up to determine the
dependence between the semantic class of a bisyllabic Chinese word and its
two characters. Simulations were performed using a three-layer
back-propagation neural net(BPNN) with various combinations of inputs. The
inputs are (1) semantic classes of the characters, (2) Entropy of the
characters and (3) semantic strengths[1] of the characters. Our results
show that we can obtain the bias of meaning class of a bisyllabic word with
an accuracy of 81% by inputting the semantic classes and strengths of the
characters. The BPNN can also be trained to predict the meaning of word to
a high precision of 83%. With the results obtained from this system, more
research on the formation of word and meaning can be performed. This system
can be applied to compound words of any languages.
Name of supervisor: Lua KT
The project title: Parsing of short sentences
IC288 or IC388 or both: both
Brief description of the project (optional):
In this project, we try to identify short sentences and parse them into
noun and verb phrases. We can then obtain direct relationship between a
noun and its verb. In LINK grammar, this is called long distance dependency
as noun and its verb can be seperated by many other words such as
adjectives, quantifiers, advers etc. In the example: He buys a kilogram of
red chilly. The noun-verb-noun becomes He-buys-chilly, taking away a
kilogram of red as attachment to the last noun chilly.
Name of supervisor: Lua KT
The project title: Information Extraction and Knowledge Discovery Through
Internet
IC288 or IC388 or both: both
Brief description of the project (optional):
Nowadays, we can obtain all sorts of information from Internet by reading
the Web pages. However, the whole process is very tiresome. It takes long
times and information are documented in a most dis-organised way. We often
spend hours or even days in the search of the information we want. In this
project, we aims to 1. Develop an Internet Agent program to do intelligent
search based a detailed specification of search objectives. 2. Analyze and
Extract the knowledge we want from the information collected. Basic AI and
NLP (natural language processing) techniques will be deployed.
Name of supervisor: Tan Tiow Seng
The project title: Real Virtual Reality Modeling
IC288 or IC288 or both: IC288 or IC388
Brief description of the project (optional):
Research and development into tools to effectively capture real
world into virtual world in computer.
Name of the supervisor: A/P A.L.Ananda
The Project Title: Investigation into the internet 2 protocol and
performance
Name of supervisors:
Dr. Ooi Beng Chin, DISCS, 874-6465, ooibc@iscs.nus.sg
Dr. Pang Hwee Hwa, ISS, 874-7859, hhpang@iss.nus.sg
Dr. Ngair Teow Hin, ISS, 874-6668, teowhin@iss.nus.sg
The project title: Mobile Java Threads
Objectives:
To extend the Java runtime environment so that threads of execution can
migrate from host to host while preserving their state information.
Motivation
In mobile computing, it is highly desirable for threads to be able to
hop between hosts. This allows a user to initiate a job on a mobile
device, and subsequently move the entire job to a server for
resource-intensive computation. It also enables an application to
roam with a user to different machines without having to shut down and
restart each time. Yet another use of mobile threads is to implement
work flow engines.
While mobility has been researched into since the advent of distributed
computing, the impact has been limited due to the difficulty of
migrating processes across platforms. However, with Java fast becoming
ubiquitous, mobility in a heterogeneous environment can finally be a
reality.
Description
The project team will implement 2 features -- checkpointing and
resumption. Checkpointing allows an application to call an API, which
will dump out a binary image of all the necessary code, program
counter(s), stack and other data structures. The resumption feature
enables the Java virtual machine to load the binary image and pick up
the execution thread(s) from the checkpoint.
The project requires the team members to develop Java classes and
probably modify the Java virtual machine. As such, the team members can
expect to be(come) conversant with Java programming and the internals of
Java.
Deliverables
- Java classes and modified virtual machine to support checkpointing and
resumption.
- A simple application to demonstrate thread migration between PC and Sun
workstations.
Name of supervisor: Dr Leow Wee Kheng (leowwk)
The project title: Keeping Track of Moving Objects in an Intelligent Room
Course code: IC288/IC388
Project description:
Motion analysis plays a key role in an intelligent environment.
For example, a security or surveillance system may keep track of the
people moving in a bank. It may detect unusual movement patterns, such
as sudden and rapid movements, which may signal the presence of robbers.
In an intelligent room, an object tracking system may keep track of the
people and objects as they move. A user who needs to locate an object
or a person can just check with the object tracking system (as in the
Star Trek senario: "Computer, where's Captain Picard?").
This project aims at developing a system that detects object movements
in an intelligent room. Having detected moving objects, it will keep track
of the current locations of the objects.A user can then query the system to find out where are the objects.
Name of supervisor: Dr Leow Wee Kheng (leowwk)
The project title: Robotic Head and Eyes
Course code: IC388
Project description:
What is a mobile robot without a head and eyes?
It would be as blind as a bat.
The objective of the project is to develop algorithms to scan for
interesting objects in a scene, and then move a robot's head and eyes
to focus on the objects. This way, the robot can selectively focus
attention at interesting things, just like humans.
Name of supervisor: Roland Yap
The project title: Extending Web Browsers
IC288 or IC388 or both: both
Brief description of the project (optional):
Web browsers like Netscape, Internet Explorer, Opera, etc. are usually
commercial software. One would like to be able to extend their functionality
in various ways but this cannot be done directly because there is no program
source and because an ideal solution should be portable among various
web browsers. Browser embedded languages like Java are one avenue but
this is limited by the Java security model.
In this project, we will look at ways of extending web browsers by using
forms of extended proxy support. For example, the user could have shortcuts
to execute queries on the web. The first part of the project will look
at building an extendable user level proxy server which can be customizable
to do various actions on the WWW by the user and also customizable behavior.
The second part of the project will look at using this approach to look
at how the user can use this to extend web pages. In partcular, how to
integrate a logic programming language such as Prolog/CLP(R) into a web
page which provides embedded computation facilities much like Java.
Name of supervisor: Ong Ghim Hwee
The project title: Fractal Image Coding
IC288 or IC388 or both: IC288
Brief description of the project (optional):
The purpose of image coding is to represent an image using as little
amount of data as possible under the condition of acceptable reconstructed
image quality. A fractal is a geometric form which has self-similar
irregular details. Techniques based on fractal properties to compress
and encode image data are called fractal image coding/compression algorithms.
These techniques make use of self-similarities existing in an image at
different resolutions under a set of affine transformations to achieve
compression. This project aims at developing various partitioning schemes
to encode images that possibly give better compression and processing time.
Name of supervisor: Ong Ghim Hwee
The project title: Multimedia Bus/MRT Route
Searching
IC288 or IC388 or both: IC288
Brief description of the project (optional):
This project is to employ multimedia tools to develop a bus/MRT route
searching system including
(a) a subsystem of image-based street maps.
(b) a subsystem for best route searching in terms of least time and/or
least cost.
(c) a subsystem that supports interactive communication schemes, allowing
users to make on-line queries.
Name of supervisor: Dr. Sung Kah Kay
The project title: Variable-pose Human Face
Detection
Subject Areas: Computer Vision, Neural Nets.
Proposed Module: IC388
Short description of the project:
In this project, we shall develop a face finding system that locates
human faces in images. We shall begin by solving a simpler problem of
locating a single (possibly non-frontal) face in a uniform image
background. If time permits, we shall use the isolated face finder to
train a robust non-frontal face finder that operates in cluttered
scenes.
Initial work will be based on an existing frontal-view face detection
approach by [Sung and Poggio 1994].
Name of Supervisor: Dr. Leong Hon Wai (leonghw) S15, 04-16
The project title: Some Applications of Clustering
Systems
Course: IC2288 or IC3288
Brief Description:
This project aims to investigate the use of clustering methods to
(i) improve network design and reliability, and (ii) obtain
clusters of ships for a container port berth allocation problem.
For the second application, clustering will be based on the
container movement matrix.
Note: In a previous research project, we have developed a general
Clustering System (in C on UNIX) that encompasses several well-known
clustering methods.
Name of Supervisor: Dr. Leong Hon Wai (leonghw) S15, 04-16
The project title: Design of Low Power Multiplexor
(MUX)
Course: IC2288 or IC3288
Brief Description:
This project deals with discrete optimization algorithms for a
binary tree decomposition problem. This problem has application
in the practical problem of low power design of digital circuits.
Specifically, we want to decompose a large n-to-1 multiplexor (MUX)
into a binary tree of 2-to-1 MUXes. The problem is to find a
decomposition that optimizes the overall power consumption.
(A paper describing our research work on this problem is
available.)
In this project, student will give *efficient* implementations of the
some heuristic algorithms for the problem.
Extensive experimental evaluation of these algorithms will be carried out.
Possible extensions of this problem will also be explored.
Name of supervisor: A/P Chua Tat Seng
The project title: Object-based Model for Image
Retrieval
Brief description:
This research is to look into higher level representation of images'
contents by using colors and regions, and explore the effectiveness of
using these higher level models for retrieval.
Name of supervisor: A/P Chua Tat Seng
The project title: Relevance Feedback in Image
Retrieval
Brief description:
Image retrieval is an iterative process. This research will look into
how content characteristics of relevant images identified may be used to
improve retrieval performance automatically. In particular, the use of
higher level content representation at obejct level will be explored.
Name of supervisor: A/P Chua Tat Seng
The Project title: Advanced Free-text retrieval
Techniques
Brief description:
This research looks into the use of addiitonal lingustic support to
improve the performance of free-text retrieval system. In particular, we
will look into using advanced digital dictionary such as WordNet to
provide additional support to word meanings and senses.
Name of supervisor: Tan Tiow Seng
The project title: Computer-Aided Road
Modeling
IC288 or IC288 or both: IC288 or IC388
Brief description of the project (optional):
Research and development into tools to effectively model city roads
for computer simulation
Name of supervisor: Tan Tiow Seng
The project title: IT in infra-structure planning---
an augmented reality application
IC288 or IC388 or both: IC288 or IC388
Brief description of the project (optional):
Research and development into augmented reality application to
planning infra-structure
Name of supervisor: Dr Leow Wee Kheng (leowwk)
Project Title: Neural Network for Controlling Robotic
Arms
Project Type: IC288 or IC388
Project Description:
Robotic manipulators (i.e., robotic arms) have been widely used
in the industry to perform routine manufacturing and assembly tasks.
However, existing robotic arms are very bulking and cumbersome,
and are, therefore, difficult to control.
Many researchers are now working on using neural networks to
control the robots. The advantage of neural networks is that they
can be trained to perform the tasks better and better.
The objective of this project is to devise a neural network that
can perform simple control tasks.
Name of supervisor: Dr Leow Wee Kheng (leowwk)
Project Title: Searching for Objects for Vision-Guided
Robotic Assembly
Project Type: IC388
Project Description:
Robotic arms have been widely used to perform routine assembly tasks.
Existing robots are blind---they can perform the tasks only if the parts
to be assembled are placed at predefined locations. This project aims
at giving the robots the sense of sight. A camera takes pictures of
the workbench on which the parts are randomly placed. A computer vision
system searches for the required parts in the image and tells the robot
where the parts are on the workbench. The robot can then pick up the
parts and assemble them. This project will focus on the searching
of the objects.
Supervisor: Dr Leow Wee Kheng (leowwk)
Project Title: Extracting Shape Features From Natural
Images
Project Type: IC288 or IC388
Project Description:
An object typically consists of several parts, each of which has a
simple shape. For example, a mug consists of a cylindrical body and a
curved handle. The shapes of these object parts can be described by
properties such as curvature, symmetry, parallelism, and convergence.
These shape features can be used in many applications such as object
recognition and retrieval of images based on shapes.
The objective of this project is to extract these shape features from
images containing natural objects and scenes.
Name of supervisor: Dr. Chan Hock Chuan
The project title: Spreadsheet Programming
IC288 or IC288 or both: either IC288 or IC388
Brief description of the project (optional):
The objective is to program, with Visual Basic for Applications
language, small auditing tools for Microsoft Excel. These auditing tools
are to help users analyse a spreadsheet. The student should be excellent
in programming.
Name of supervisor: Feng Ming Dong
The project title: Multithreaded Programming System on
AP3000
Brief description:
Multithreading is widely available in almost all operating systems.
Multithreaded programming languages have been used in many commercial
applications. For example, Netscape uses threads to support JAVA
applets. There is an interesting runtime system for a multithreaded
programming language Cilk (see more details of Cilk at
http://theory.lcs.mit.edu/~cilk). The Cilk runtime system features
provably efficient scheduling and virtual shared memory support. This
project studies porting of the Cilk runtime system onto the
department's own MPP machine AP3000.
Name of Supervisor: Dr. Leong Hon Wai (leonghw) S15, 04-16
Project Title: Vessel Packing using Mathematical
Programming
Course: IC2288 or IC3288
Brief Description:
In this project, we study the use of mathematical programming methods
in solving the problem of "packing" vessels into a section of the harbour.
This problem arises as a sub-problem of a larger R&D project on
berth allocation of ships in a container port.
We are given a collection of ships (vessels) that have to be berthed
in a section (of the harbour) consisting of a set of berths.
We are also given all other pertinent information such as the lengths,
the arrival and departure times, and the berthing times (length of time
that a vessel has to be berth) of the vessels.
We want to assign the vessels to berths (long vessels may occupy more
than one berth) so as to satisfy (as much as possible) the departure
times of the vessels.
Name of Supervisor: Dr. Leong Hon Wai (leonghw) S15, 04-16
Project Title: Search Techniques for the Berth
Allocation Problem
Course: IC2288 or IC3288
Brief Description:
In this project, we study the berth allocation problem (BAP).
This problem arises as a sub-problem of a larger R&D project
investigating the BAP.
The BAP is described as follows:
Vessels arriving at a container port are berthed in a section of a
wharf, and the containers they ferry are then transferred to the port
or to other vessels. We are interested in finding good berth allocations
so as to minimize the overall movement of the containers.
We currently have one solution to the BAP that uses genetic algorithm.
In this project, we want to explore the use of other search techniques
such as simulated annealing and tabu search.
Name of Supervisor: Dr. Leong Hon Wai (leonghw) S15, 04-16
Project Title: The Airport Gate Assignment
Problem
Course: IC2288 or IC3288
Brief Description:
In this project, we study the Gate Assignment Problem (GAP)
of allocating gates to planes arriving at the airport terminal.
This problem arises as a sub-problem of a larger R&D project
investigating the GAP and related problems.
The GAP is described as follows:
Planes arriving at an airport are assigned gates in the terminals
of the airport and the passengers either go into the airport or
are transferred (connected) to other flights. We are interested
in finding good gate assignments so as to optimize resources and
minimize the overall passenger movement.
We currently have a solution to a related problem, the BAP (see
my other project proposal). In this project, we want to modify
the BAP solution for solving the GAP and possibly explore other
solution methods.
Name of supervisor : Dr Chin Wei-Ngan, Dr Khoo Siau Cheng
The project title : NESL - A Functional Language for
Nested Parallelism
Brief Description:
NESL is a simple functional language which supports
portable nested data parallelism. It has a simple
programming model and is portable across a range of
parallel and serial machines. Due to its simplicity
and expressiveness, the language has been found to be
an excellent tool for both teaching and writing parallel
programs.
The aim of this project is to port the NESL
language to the AP3000 parallel systems. The focus
will be on ensuring that a core library of data-parallel
operations are efficiently supported in the port.
Name of supervisor: Dr Khoo Siau Cheng (khoosc@iscs.nus.sg
7726730
S15 #05-10)
The project title: Algebra of Programming
Description of the scope of the Project :
An algebraic approach to programming is suitable both for the
derivation of individual programs and for the study of programming
principles in general. Such an approach is usually called "calculating
programs". Recently, there have been breakthroughs in using this
approach, and researchers have found various ways of describing
program transformations based on the algebra of programming. Some of
these program transformations can even be automated and used in
compiler design.
All these program transformations, and the algebra of programming in
general, are based on a mathematical framework, known as the
"categorical calculus", which is known as the "category theory" in the
field of mathematics. With the help of this calculus, the standard
principles of algorithm design (and transformation strategy) can be
formulated as theorems about classes of problems whose specifications
possess a particular structure. These theorems say that, under
appropriate conditions, a certain strategy (eg: transformation) works
and leads to a particular form of abstract solution (eg: more
efficient programs).
The scope of this project is :
0. to understand category theory;
1. to investigate and understand the categorical approach to
programming;
2. to investigate the use of such approach in performing a particular
program transformation (eg : fusion, tupling, etc.) ;
3. to implement such program transformation.
Name of supervisor: Andrew Lim Leong Chye (alim@iscs.nus.sg)
The project title: 3-D Container Packing
Problem
Name of supervisor: Andrew Lim Leong Chye (alim@iscs.nus.sg)
The project title: Ship Berthing Problem with
Delays
Name of supervisor: Andrew Lim Leong Chye (alim@iscs.nus.sg)
The project title: Vehicle Routing Problem
Name of supervisor: Liu Bing (liub@iscs.nus.sg)
The project title: Discovering interesting knowledge
in databases
Name of supervisor: Yeo Gee Kin (yeogk@iscs.nus.sg)
The project title: Model Queries
Brief description:
Query in the context of modeling concerns not only of data, but
also of model structure, its entities and attributes. Because of
different model views, different representations and therefore
structures are used in a modeling system. Providing answers to model
queries very often involve moving between structures, integrating
representations and scheduling and executing processes. In particular,
solution query has been used broadly to refer to requests for any
model manipulation performed by solvers and is not limited only to
retrievals, as in the database sense, but also includes evaluation,
satisfaction and optimisation. This project will review works done
in model queries and identify viable approaches for implementation
in a modeling system prototype,
VMS/SM(http://www.iscs.nus.sg/~yeogk/MM/research/vms_sm).
Name of supervisor: Yeo Gee Kin (yeogk@iscs.nus.sg)
The project title: Multi-lingual support on an Internet
simulation game
Brief Description:
MAGNUS (http://www.iscs.nus.sg/~yeogk/MAGNUS/magnus.html), a
simulation game used to support a course in decision support system
in DISCS are being internationlised to be available on the WWW.
(http://www.iscs.nus.sg/~stfpool/magnus). This project will re-
examine its multi-lingual support to allow its use on multi-cutural
settings. An experimental multi-lingual application support
environment will be used to generate the required multi-lingual
support modules.
Name of supervisor: Dr. Sung Kah Kay
The Project Title: Parameterizing face structure and texture for face
recognition
Subject Areas: Computer vision, Face recognition.
Proposed Module: IC388
Short description of the project:
Over the past few years, the eigenspace approach for face recognition
[Turk and Pentland 91] has received a significant amount of attention
within the face recognition research community. The approach aligns a
face (manually) to a canoncal grid, and reconstructs the aligned face
image as a linear combination of several pre-stored face images
(a.k.a. basis functions). The linear combination coefficients are used
as features for face recognition.
In this project, we shall develop the front-end of a face recognition
approach that uses the structure (i.e. the original shape and precise
location of facial features) of a face as additional features for face
recognition. We shall develop a parameterized model for face structure,
similar in spirit to the eigenspace approach for parameterizing face
texture.
Given a face image, the front-end procedure you develop will attempt to
align the face with a canonical face structural model, and recover a set
of alignment coefficients as recognition features.
Name of supervisor: Dr. Sung Kah Kay
The Project Title: Training a Vision-based Mobile Robot Navigation
Controller
Subject Areas: Computer Vision, Neural Nets.
Proposed Module: IC388
Short description of the project:
For a mobile robot to be truly autonomous, it must be able to move
along a passage way without running into obstacles. In this project,
you will train and implement a vision-based mobile robot control
algorithm to perform some simple indoor navigation tasks.
We shall look at ways of automating the training process by using sonar
sensors on the robot to train its visual guidance system.
Our goal is to produce a scaled-down indoor version of NavLab: the CMU
AGV guidance system.
Name of supervisor: A/P Tan Chew Lim
The project title: GENETIC PROGRAMMING OF NEURAL LOGIC
NETWORK
Brief description:
Neural Logic Network is a kind of hybrid between neural network
and rule-based expert system. It has the ability to encode
knowledge like an expert system and yet at the same time it can
learn by allowing the weights in its rule networks to change,
thru some neural network learning algorithms such as
backpropagation. The present project is to study the possibility
of applying genetic programming on neural logic network as
another form of learning.
Genetic programming is an extension of the conventional genetic
algorithm in which each individual in the population is a
computer program. In the case of neural logic network, each
computer program to be subject to genetic programming will be the
basic rule network. Thru genetic programming, it is hoped that
the system will be able to learn to construct an entire neural
logic network on its own.
A neural logic network environment has already been developed.
The student will first study the behavior of the nerual logic
network within the environment and then move on to experiment the
genetic programming.
Name of supervisor: Tan Kian Lee
The Project Title: Distributed Spatial Join
Processing
IC288 or IC388 or both: IC2288
Brief:
In a distributed environment, a spatial join may involve
2 relations that are stored at two different sites. As
communication may be expensive especially for spatial data
which are usually large in size, new spatial join algorithms
are needed. This project will design and study new spatial
join processing algorithms for distributed environment.
Name of supervisor: Ang Chuan Heng
The project title: Recursive linear hashing and its
improvement
Brief description:
In this project, student will read a paper, implement the recursive linear
hashing algorithm, explore various ways to improve it and carry out some
experiments to compare the performance of the variants.
Name of supervisor: Ang Chuan Heng
The project title: Filtered spiral hashing
Brief description:
In this project, student will read at least 2 papers, implement the spiral
hashing algorithm and filtered spiral hashing, and carry out some
experiments to compare the performance of the variants.
Name of supervisor: Ang Chuan Heng
The project title: Linear hashing and its
improvements
Brief description:
In this project, student will read a paper, implement the linear
hashing algorithm, explore various ways to improve it and carry out some
experiments to compare the performance of the variants.
Name of supervisor: Dr. Wang Ke (wangk@iscs.nus.sg) S16, 06-13
The Project Title: Data warehouse keeper for Knowledge discovery and data
mining
Brief Description:
Knowledge discovery and data mining is becoming
a key IT technology for the success of business and industry.
It helps an organization to make sensible decision through
intensive data analysis. Today data mining is a multi-billion dollar
industry --- very easy to find jobs. The crucial component of data mining
is the data --- one has to know where to find the data before mining
any "gold" from it. In this project, the student will search and
investigate what data sources exist on the internet and WWW. For
different discovery tasks, data can have different formats.
Some are relational tables in a database (most business types), some
are HTML files on the web, some are transactional data (what are bought
in each transactions) in the supermarket case, some are image/sound data,
some are textual data (like news), some are sequences data (like
weather and stock data), etc. In many cases, data have to be interpreted
using background or domain knowledge. The student will first
get familar with all types of knowledge discovery and data mining tasks.
Then he/she will search and classify data sources existing
on internet or web according to use for different tasks. This
project is particularly suitable for students who wish to
do research on knowledge discovery and data mining in later stages
such as Honour, Master or Ph.D
Name of supervisor: Tan Kian Lee
The project title: Spatial Join Processing by Partial
Precomputations
Name of supervisor: Tan Kian Lee
The project title: Query Processing and Optimization in Heterogeneous
Database Systems
Name of Supervisor : Dr Khoo Siau Cheng
The Project Title : Super-linear Speedup via Tupling
Transformation
Brief Description :
Source-to-source program transformation is the bridge between
easy-to-read programs and their efficient-to-execute counterprats.
Tupling transformation is distinctive from other existing program
transformation in that it realises super-linear speedup in the
transformed program. Candidate (programs) for tupling transformations
abound, like fibonacci function, tower of hanoi, etc. All these
candidates share a common syntactic property that enables tupling
transformation to take effect: the property of having redundant
computation.
In this project, the student will learn the technique of tupling
transformation, the method to detect redundant computation in a program.
His/her task is then to implement the transformation and collect some
statistics on its effectiveness in achieving super-linear speedup.
Name of Supervisor: Ang Chuan Heng
The project title: S-trees vs R-trees
Brief Description:
In this project, the student has to read afew papers, implement S-tree,
R-tree, and R*-tree, and carry out some experiemtns to compare their
performance.
Name of Supervisor: Teo Yong Meng
The project title: A Study of Distributed Interactive
Simulation
Name of Supervisor: Teo Yong Meng
The project title: Strategies for Distributed Memory
Management
Name of Supervisor: Teo Yong Meng
The project title: Scenario Analysis on a Parallel
Computer
Name of supervisor: Dr Lucas Hui
The Project Title: Authentication Scheme based on
Question-Answering
Brief description:
User authentication on a computer network such as internet is an
important issue for successful deployment of information technology.
Popular solutions include the use of passwords, public key
cryptosystems, secure hardware like smart cards, or biometric means
such as fingerprint or voice recognition. Recently, there is a new
approach for network user authentication. In this kind of schemes,
when a human user wants to identify himself to another party such as a
host computer over the network, the other party will generate a set of
questions for the user to answer. The core idea of the scheme is that
the question-answering model is specially designed in a way that enough
information content are reviewed in the answers to identify the user,
but yet an attacker who wiretap the communication channel cannot obtain
enough information to impersonate the user.
The aim of this research project is to study the properties different
question-answering models, as well as to study the values of these
models in practical settings such as a customer authenticating himself
in front of a banking machine, or students authenticating themselves in
an electronic examination system. Research work can be either
analytical, experimental, or both.