Tomohiro Odaka

A study of learning splice sites of DNA sequence by neural networks

To predict of splice sites in DNA sequence, we developed a neural network system with back propagation. This system has a flexible network definition language which can describe any network structure. Three types of neural network were defined using the system for the prediction of splice sites. The neural networks are trained by the arrangements of bases around the splice sites of DNA sequences. The results of simulation showed the excellent ability of the neural networks to predict splice sites by applying and testing the arrangements of DNA sequences. This system also were used to predict the effects of point mutations on the splicing of the IX factor gene which may cause hereditary disease.

The effect of a multiple literature database search: a numerical evaluation in the domain of Japanese life science

In literature database searching, we show that it is necessary to use plural databases for a more improved search. We also compare the results of a single database search with that of multiple database search in the domain of Japanese life sciences. We searched the MEDLINE and EMBASE using the same search terms. There were some differences in the results, owing to differences in the journals and recording methods. We herein show some of the differences in the journals contained in both databases. Furthermore, we show the differences in the number of papers derived from the same journal. Next, as an example of a practical search, we selected some universities in Japan, searched both databases regarding papers published from these universities and then merged the results by hand. According to our results, only 63% of all papers were common to both databases.

A random allocation system with the minimization method for multi-institutional clinical trials

This paper describes the random allocation system used to perform precise and rapid treatment assignments in multi-institutional clinical trials. This system is based on sophisticated randomization procedures, according to Pocock and Simons minimization method and Zelens method for institution balancing. The major advantage of randomized treatment assignments with this system is to balance treatment numbers for each level of various prognostic factors over the entire trial and at the same time balance the allocation of treatments within an institution. Therefore, the randomized treatment assignments by this system can prevent degrading of the statistical power of a particular treatment factor. This system is designed to run on a small-sized notebook computer and therefore can be set up beside a telephone for registration, without occupying a large space. At present, this system is conveniently being used in two clinical trials.

A hybrid command sequence model for anomaly detection

A new anomaly detection method based on models of user behavior at the command level is proposed as an intrusion detection technique. The hybrid command sequence (HCS) model is trained from historical session data by a genetic algorithm, and then it is used as the criterion in verifying observed behavior. The proposed model considers the occurrence of multiple command sequence fragments in a single session, so that it could recognize nonsequential patterns. Experiment results demonstrate an anomaly detection rate of higher than 90%, comparable to other statistical methods and 10% higher than the original command sequence model.

Fuzzy linguistic facial caricature drawing system

Considers a realization of a process drawing facial caricatures and exaggerating them using the concept of fuzzy information granulation. Face space is introduced, for the facial caricature drawn in computer graphics. The linguistic terms which indicate features of ones face are given as a fuzzy set by means of the concept of computing with words. Exaggerated knowledge is realized as rules on linguistic labels. Also simulation examples of facial caricature drawing and exaggerating are shown.

A general purpose neural network simulator system for medical data processing

We developed a general purpose neural network simulator system for medical data processing. This system has a flexible network definition language. Users can define arbitrary hierarchical neural networks using the definition language to analyze medical data that contains some complex patterns. The learning algorithm used in this system is back propagation. Learning curves are displayed on multiple windows. This is a general purpose system, so it can be used for various kinds of medical data processing such as one dimensional signal processing or two dimensional image processing. The system can run on a standard UNIX workstation, which is faster and more powerful than most personal computers. The system needs an X window system/Motif Motif and C compiler. These are standard system programs already available on most UNIX workstations. The source code of the system can be retrieved from our anonymous ftp site via Internet.

A flexible random allocation program for multi-institutional clinical trials

This paper describes a flexible random allocation program that assigns treatments to patients according to their prognostic factors in multi-institutional clinical trials. The source lists are available in the appendix of this paper. This program is based on Pocock and Simons minimization method and Zelens method for institution balancing. The numbers of institutions, treatments, and prognostic factors can be set arbitrarily. The maximum number of institutions, treatments, or prognostic factors that can be accommodated by the program is limited only by the size of the main memory. For example, an IBM-PC with a 640KB main memory can run a program of 1500 institutions, 4 treatments and 20 prognostic factors.

Refining fuzzy control rules for the inverted double pendulum system based on a transformation-type genetic algorithm

We have been studying fuzzy control of the inverted double pendulum in computer simulations. We can control the inverted double pendulum by fuzzy control rules acquired by trial and error. In this research, we attempted to acquire fuzzy control knowledge for the double pendulum automatically by using a genetic algorithm (GA). In order to adjust the rules of attitude and position control of the double pendulum by the GA, we regarded rule acquisition as a multiobjective optimization problem and formulated hierarchical control knowledge. This paper proposes a new method of treating hierarchical control knowledge by expanding the GA with a local improvement mechanism. This method, based on genetic transformation in microorganisms, allows us to acquire better fuzzy control knowledge of the double pendulum.

A visual data analysis system for the medical image processing

We developed a visual data analysis system that can easily manage a large volume of medical imaging data. This system can analyze sets of imaging data using general image processing methods, so that various kinds of medical imaging data such as ECG charts, X ray image films, and MRI images, can be processed. The system has a graphical user interface (GUI). A physician who is novice at the system can manipulate the imaging data intuitively by pull down menus, pop up menus and buttons within the window system. The system can run on a standard UNIX workstation which is faster and more powerful than most personal computers. The system needs an X window system/Motif and C compiler. These are standard system programs already available on most UNIX workstations. The source code of the system can be retrieved from our anonymous ftp site via Internet.

Micro-mainframe-link personal clinical research system

We constructed a micro-mainframe-link clinical research system for personal use (Personal Clinical Research System). This system was developed with both a mainframe computer and a personal computer (PC). The prepared programs included a database manager (on the mainframe computer), a user interface program (on the PC), and a communication control program that connected the mainframe computer with the PC. The database on the mainframe computer was constructed by two methods. The first method was to transmit data from the PC to the mainframe computer. The second method was to extract data from the patient information database. Using this system, a physician is able to construct a personal research database that contains interesting data for the physician. In addition, the physician is able to accumulate data on a special field using this system. A discharge summary system is now in operation as an example of this system.