Page 141 - My FlipBook
P. 141
Brochure 2020
researchers have resorted to heuristic, randomized, or V. Graph Theory and Algorithms
approximation algorithms to nd near-optimal solutions.
In this field, uncapacitated and capacitated facility Graphs are used to model many important applications
location problems are among the most central research and are the main tool for solving many theoretical
topics. These problems have played important roles over problems. We often start by probing fundamental
several decades in the development of approximation theoretical problems, such as the structures of graphs,
algorithms due to the simplicity and elegance of their with certain properties. With these properties, we can
problem models and the diversity revealed in their then design efficient solutions. We are working on
abstract linear programming formulations. Our goal is e cient graph algorithms based on streaming models.
to tackle these kinds of facility location problems from
a di erent perspective based on our previous research. VI. Computational Learning Theory
Throughout the process of algorithmic research, we aim
to develop not only efficient algorithms and suitable Many situations in daily life require us to make repeated
approximation solutions for these problems, but also decisions before knowing the resulting outcomes. Such
endeavor to derive fundamental techniques and tools scenarios have motivated study of the so-called online
that can be used for a wide range of related problems in decision problem, in which one must iteratively choose
combinatorial optimization. an action and then experience some corresponding
loss for a number of rounds. To investigate this
IV. Massive Data problem, we are identifying natural scenarios in which
online algorithms with improved performances can
• Efficient Data Intensive Algorithms: Given rapid be designed. Moreover, we have discovered new
developments of computer and communication applications of this problem in diverse areas, such as
technology, it has become much easier to access or machine learning, game theory, and complexity theory.
store massive amounts of data electronically. We are
interested in research problems concerning e cient VII. Robotics
computation of massive data, which include e cient
epidemic simulation, visualization and construction Planning smooth paths is a fundamental task for
of disease networks, and classical computer games. applications of unmanned vehicles such as service
robots, self-driving vehicles and unmanned aerial
• Logics for Massive Data: Considerable amounts of vehicles, necessitating path and operation constraints
information and knowledge are implicit in massive for safety. We have studied different methodologies
data. We intend to study the problems of knowledge and variants of path primitives for smooth curvature-
representation and reasoning in data science by using bounded path generation in scenarios such as lane
formal logic. With proper representation frameworks chang and state-to-state transfer with minimized travel
and logical formalisms, knowledge discovered from time or path length as the criterion.
massive data can be used in data-intensive intelligent
systems.
139
researchers have resorted to heuristic, randomized, or V. Graph Theory and Algorithms
approximation algorithms to nd near-optimal solutions.
In this field, uncapacitated and capacitated facility Graphs are used to model many important applications
location problems are among the most central research and are the main tool for solving many theoretical
topics. These problems have played important roles over problems. We often start by probing fundamental
several decades in the development of approximation theoretical problems, such as the structures of graphs,
algorithms due to the simplicity and elegance of their with certain properties. With these properties, we can
problem models and the diversity revealed in their then design efficient solutions. We are working on
abstract linear programming formulations. Our goal is e cient graph algorithms based on streaming models.
to tackle these kinds of facility location problems from
a di erent perspective based on our previous research. VI. Computational Learning Theory
Throughout the process of algorithmic research, we aim
to develop not only efficient algorithms and suitable Many situations in daily life require us to make repeated
approximation solutions for these problems, but also decisions before knowing the resulting outcomes. Such
endeavor to derive fundamental techniques and tools scenarios have motivated study of the so-called online
that can be used for a wide range of related problems in decision problem, in which one must iteratively choose
combinatorial optimization. an action and then experience some corresponding
loss for a number of rounds. To investigate this
IV. Massive Data problem, we are identifying natural scenarios in which
online algorithms with improved performances can
• Efficient Data Intensive Algorithms: Given rapid be designed. Moreover, we have discovered new
developments of computer and communication applications of this problem in diverse areas, such as
technology, it has become much easier to access or machine learning, game theory, and complexity theory.
store massive amounts of data electronically. We are
interested in research problems concerning e cient VII. Robotics
computation of massive data, which include e cient
epidemic simulation, visualization and construction Planning smooth paths is a fundamental task for
of disease networks, and classical computer games. applications of unmanned vehicles such as service
robots, self-driving vehicles and unmanned aerial
• Logics for Massive Data: Considerable amounts of vehicles, necessitating path and operation constraints
information and knowledge are implicit in massive for safety. We have studied different methodologies
data. We intend to study the problems of knowledge and variants of path primitives for smooth curvature-
representation and reasoning in data science by using bounded path generation in scenarios such as lane
formal logic. With proper representation frameworks chang and state-to-state transfer with minimized travel
and logical formalisms, knowledge discovered from time or path length as the criterion.
massive data can be used in data-intensive intelligent
systems.
139
