Statements on Research


Human beings have accumulated a great deal of knowledge on their environment and yet know very little about the mechanisms that allow them to understand. I believe the next frontier of human endeavor is to gain the knowledge on the working of our brain and our mind. Building an intelligent machine aims at our endeavor in two ways. First, the intelligent machine may reflect functions of our brain and our mind. Second, it may help us conduct further research. In the quest for building an intelligent machine, I conduct research on both building the brain and developing the mind of a machine.


I conduct research in parallel and distributed computing hardware because I believe that the sequential computers of today are inadequate to be the brain of an intelligent machine. I began my research from the ground up by proposing parallel memories. I conceived and designed a special type of memory called general associative memory. The memory is not only able to record data but is able to search the data in parallel. Next, I proposed a special type of parallel computer architecture. The architecture consists of many processing elements each of which has distinct processing functions just like each part of our brain has specific functions. Much more research is needed for building the brain of an intelligent machine. I intend to continue the search. 


I also conduct research in machine learning because I believe that learning is an essential function for an intelligent machine. I conceived and implemented a learning machine that is capable of learning both combinational and sequential tasks. The machine learns from examples and by observations just like human often learns by imitation. One important feature of the machine is its ability to learn the orders and dependences of events. Thus the machine is capable of learning to perform tasks that consist of sequences of steps. The machine is also capable of adapting to change and is capable of handing inconsistent information. Much more research is needed for developing the mind of an intelligent machine. I aim to continue the quest.


FIT Group



Research Experience


I grow through research. Building electronic devices and writing computer programs have been my hobbies since I was a teenage, I grew up through research in hardware and software. Before I finished high school, I was already a certified electronic technician and had already completed an Artificial Intelligent game, which I called Master+Mind.  


I continued researching in both hardware and software through graduate school. For hardware I proposed and developed Artificial Intelligent machines as documented in my master thesis “Parallel Distributed Computer Architecture and General Associative Memory for Artificial Intelligent Processing” and used the “Generic Associative Memory for Information Retrieval”. For software I proposed and developed a new Machine Learning model as documented in my Ph.D. dissertation “Automata for Learning Sequential Tasks”.


As a professor, I continue my quest in researching in both hardware and software. For hardware, I proposed a new system for “Applying Learning by Example for Digital Design Automation” (Choi 2002). For parallel processing, I proposed a new framework for applying Intelligent Agent Technology for parallel and distributed computing as described in “Distributed Object Space Cluster Architecture for Search Engines” and “Agent Space Architecture for Search Engines”. In addition, I investigated using “Component-Based Distributed Computing for Numerical Simulation”.


I continue my work on Machine Learning by relating inductive inference to information compression as documented on “Inductive Inference by Using Information Compression”. Then, I decided to apply my experience on Artificial Intelligence and Machine Learning for “Making Sense of Search Results by Automatic Web-page Classifications” (Choi 2001). I set up a large research group consisting up to fifteen students conducting research on Internet related applications in the areas of Web page classification and clustering, search engine technology, and Information Technology.


For web page classification and clustering, my research group focused on organizing web pages into hierarchies of meaningful categories. We proposed algorithms for web page classification by using single-path search from the root to a leaf of a category tree and by automatically adding new categories into the tree when needed (see “Automatic Web Page Classification in a Dynamic and Hierarchical Way” and “Dynamic and Hierarchical Classification of Web Pages”). We also proposed to use hyperlinks on a web page by considering linked pages to be semantically related as described in “Applying Semantic Links for Classifying Web Pages”. To address the problem when predefined categories are not available and thus classification methods are not applicable, we proposed clustering algorithms to automatically organize web pages into a hierarchy of groups as described in “Bidirectional Hierarchical Clustering for Web Mining”. To address the problem of labeling or naming each newly clustered group, we proposed techniques for automatically summarizing web pages by using keywords and synonyms (“Abstracting Keywords from Hypertext Documents”). For further details on the state of the art techniques and subsystems used to build automatic web pages classification systems, see our book chapter on “Web Page Classification”.


For search engine technology, my research group focused on incorporating our web page classification and clustering technologies into our search engine innovations creating a next generation of Information Classification and Search Engine (iCSe) system. We also addressed the problem of “Speeding up Keyword Search for Search Engine” by proposing new schemes for organizing and searching keywords. In addition, we utilized our parallel and distributed architecture for building search engines that can support millions of users and can search billions of web pages (“Distributed Object Space Cluster Architecture for Search Engines” and “Agent Space Architecture for Search Engines”).


For Information Technology, besides working on our Information Classification and Search Engine system, we proposed new methods for “Network Traffic Reduction by Hypertext Compression” and “An Adaptive Web Cache Access Predictor Using Neural Network”.

Moreover, I also have other research interests including, Robotics and Virtual Reality (“Humanoid Motion Description Language”) and Biomedical applications (“Text to Fingerspelling and Speech on the Amiga Micro-computer”). 


Besides the above highlights of some of my published works, my current research under development and publications under review are not shown here. I grow through research and which will not stop. The future of computing is moving from individual processing units to communities of self organizing agents. I aim to continue the quest.


Switching from scholarly research to reality, research may also mean bringing in external funds for a university. To this end, I helped found a new research center bringing in millions of dollars each year. In return, the center has been supporting my research group. Our new research center called Center for Entrepreneurship and Information Technology focuses on innovations in Information Technology and on applying new technologies to simulate the growth of high-tech industries in the State. To this end, I submitted four reports of inventions to our university, completed two patent applications (“Method and Apparatus for Information Classification & Search Engine Design” and “Method and Apparatus for Organizing Information on Internet”), developed a fully working prototype Information Classification and Search Engine (iCSe), and negotiated with several potential customers for licensing and deploying our technologies. As demonstrated by Google, which has estimated worth of $23 billion dollars, our new information classification and search engine technologies have potential for bringing in large external funds for our university. We should continue the entrepreneurship.