Research Activities of the Students
Researches of Graduate Students
- Name: Kyohei Watarai
- Title: Decision boundary learning and its application to inducing high performance neural networks
- Abstract: The goal of this research is to design a multimedia
analyzer (MA) that can be embedded in portable devices. This MA can
recognize different multimedia (e.g. text and image) patterns and
help the user to analyze the multimedia contents more
efficiently. To realize the MA in an environment with limited
computing resource, we propose a new concept called decision
boundary learning (DBL). The basic idea is to generate training
patterns close to the decision boundary (DB), so that a neural
network (NN) with high generalization ability can be obtained.
Experimental results obtained so far show that the NNs so obtained
are comparative in performance to the SVMs although the former are
much more compact.
- Name: Yu-Cheng Lin
- Title: A study on standardization of ontology design based on decision rules
- Abstract: Ontology is a semantic technology. It can be used to
describe knowledge and resources. In recent years, ontology-based
systems are becoming very popular. They have been used in different
domains. An ontology-based system consists of an ontology model, an
inference engine, and a rule base. In this study, we try to
standardize the process of designing ontology systems, so that
different uses can obtain the same system for solving the same
problem, and the system does not contain useless components. The basic
idea is to generate decision rules based on context information, and
select useful features for making decisions.
- Name: Shun Endou
- Title (tentative): Development of an iPad Application for Generating JJ1017 Code
- Abstract: (under construction)
- Name: Yuutaro Minakawa
- Title (tentative): A study on multimedia retrieval based on support vector machine
- Abstract: The purpose of this research is to develop a multimedia
retrieval system that can help users to find useful contents given a
reference content x. Here, x can be a keyword list, a txt-file, an
image-file, or a sound-file. We try to extract "word-features" from
different types of contents, and then use the same machine learner to
make decisions. Results will be reported in near future.
- Name: Yuya Kaneda
- Title (tentative): A study on portable device based user modeling
- Abstract: User modeling is important for providing personalized
services. So far, many modeling methods have been studied and used in
information retrieval systems, in contents recommending systems, and
so on. However, personalized services can be conflict with the user
privacy. One way to solve the problem is to build the user model in
his/her own computer, and he/she can use this model to filter out
useless contents locally and make the service system un-aware of
his/her preference. To make this kind of "local personalized service"
availabe anywhere and anytime, the user model should be implementable
in portable devices. For this purpose, we must solve problems related
to data, feature, and dimension reduction. We also need to propose
learning models that are powerful enough to cover different domains.
Graduation Researches
- Yoshiharu Nanaumi: Automatic determination of feature points for image morphing
- Shigeki Okabe: Finding trustable friends in SNS based on machine learning
- Fumiya Nagashima: Generating natural facial images based on IDE and image morphing
- Kazuho Kanno: Image recognition in smart phones
You can find the research titles of the students already graduated from our laboratory by clicking
backnumbers.