|Affiliation||Department of Computer Science and Engineering|
|Fields of Research||Multimedia information retrieval and data management / Automatic annotation to multimedia / 3D shape retrieval / Data mining / Text mining / Web mining|
|Degree||Ph.D. (Rensselaer Polytechnic Institute, USA)|
|Academic Societies||ACM / IEEE / IPSJ / IEICE / NLP / JSAI / Japan Database Society|
Please append "tut.jp" to the end of the address above.
|Laboratory website URL||http://www.kde.cs.tut.ac.jp/|
The most recent research has been focused on 3D shape retrieval (from 3F query or from 2D images/sketches).
We are boasting of the world best search performance lately with our patented technologies.
Research interests include multimedia (text, image, and video) retrieval, feature extraction, feature selection, classification, and automatic annotation.
Also, we have investigated time series data mining technologies.
A particular application we have done so far includes the prediction of tomato yields (crops) from the past environmental data as well as past growth data.
In addition, Web mining, patent mining, and text mining (including sentiment analysis) are now under our focused areas of research.
Theme1：Research on 3D Shape Retrieval, Shape Classification, Segmentation, and Automatic Annotation based on Supervised Learning
3D shape models are in great use in a variety of industries ranging from mechanical CAD, flight simulator, architecture, civil engineering to game and entertainment industries, education, and commercial films, and 3D movies. However, it is formidable to create each new 3D model from scratch, because making a 3D shape models is very time consuming task. To overcome these difficulties, we have developed a series of new technologies (where some of them are patented) to retrieve 3D shape objects with remarkable accuracy. In 2013 and 2014, our technology is proved to be the world number 1 in search performance of several competition tracks in SHREC2013 and SHREC2014, respectively.
Theme2：Time Series Data Minig
Given multi-modal sensors, or any sensors that can keep time series data, we are conducting research on data mining from a vast collection of time series data. In the past, we have succeeded in predicting "Tomato" yields (crops) from the past 3 years' data (both environmental data and growth data) with 70-80% accuracy.
With the increase of user opinions, twitters, and blogs on the web, the interest in mining sentiment has grown rapidly.
In this research, given a text document, we are investigating the way to estimate the polarity (positiveness, negativeness, and neutrality) of the words, the sentences, and the phrases appearing in the document.
We are also attempting to estimate the polarity of the document as a whole.
Title of class
Computer Programming B13610020 Compulsory (B3)
Introduction to Computer Architecture, B13530090 (B2)
Advanced Data Mining, M23630080 (M1)