Misc - NAKAMURA Takayuki

機械学習法のロボット知能化システムへの応用(2)

中村 恭之, 和田 俊和

機械の研究 ( 養賢堂 )  58 ( 2 ) 263 - 267   2006.02

機械学習法のロボット知能化システムへの応用(1)

中村 恭之, 和田 俊和

機械の研究 ( 養賢堂 )  58 ( 1 ) 7 - 16   2006.01

2A1-E08 Tightly coupled planning and model learning for real robot behavior learning

Kawahara Terumi, Nakamura Takayuki, Wada Toshikazu

ロボティクス・メカトロニクス講演会講演概要集 ( 一般社団法人日本機械学会 )  2006   "2A1 - E08(1)"-"2A1-E08(4)"   2006

　View Summary

This paper presents a new framework for learning dynamic robot behavior such that a wheeled mobile robot moves to the destination with wheel slippage. First, our method learns the body dynamics from a large number of sensory-motor instances using PaLM-tree algorithm, which approximates nonlinear mapping in arbitrary preciseness. The mapping specifies a transition model from state-action pair to next state. An optimal action sequence generating the transition from initial to goal states is obtained by the branch-and-bound algorithm using the estimated transition model. Then, new sensory-moto...

Instance Based Construction of Distinctiveness Map and Its Application : Distinctiveness Map Based Stereo Tracking

IIZUKA Takeo, WADA Toshikazu, HUA Chunsheng, NAKAMURA Takayuki

IPSJ SIG Notes. CVIM ( 一般社団法人情報処理学会 )  2005 ( 112 ) 9 - 16   2005.11

　View Summary

This paper presents a measure for color based target detection and tracking. The measure, distinctiveness, extends our nearest neighbor classification based color target detection to target tracking. Since the color target detection can only detects target regions, if no target is detected near tracking region, the tracking fails. However, if we have a measure representing the distinctiveness of the target color, we can keep tracking by finding the regions having higher distinctiveness. The distinctiveness is defined not only by the color distance from the target color set but also by the d...

事例を用いた顕著性マップの構築とその応用

飯塚,和田,華,中村

情報処理学会研究報告2005-CVIM-151     65-70   2005.11

Learning Nonlinear Mapping based on Recursive Partitioning of Data Space : Current and Old Trend on Regression Tree

NAKAMURA Takayuki, WADA Toshikazu

Journal of Information Processing Society of Japan ( 一般社団法人情報処理学会 )  46 ( 9 ) 1030 - 1038   2005.09

LF-006 Estimating the Number of Tracking Targets for Human Location Tracking Using Floor Pressure Sensors

Satoh Tetsu, Wada Toshikazu, Nakamura Takayuki

情報科学技術レターズ ( Forum on Information Technology )  4 ( 4 ) 103 - 104   2005.08

High Performance Control of Active Camera Head Using PaLM-Tree

T.Nakamura, Y.Sakata, T.Wada and H. Wu

IEEE/ASME International Conference on Advanced Intelligent Mechatronics 2005     963-968   2005.08  [Refereed]

　View Summary

基本アルゴリズムの考案,実験, 検証

MDL基準に基づく区分的関数あてはめによる写像学習

上江洲吉美, 中村恭之, 和田俊和

画像の認識・理解シンポジウム(MIRU2005)     144 - 150   2005.07  [Refereed]

視点固定型パン・チルトステレオカメラによるリアルタイム3次元位置計測システム

飯塚健男, 和田俊和, 中村恭之, 加藤丈和, 吉岡悠一

画像の認識・理解シンポジウム(MIRU2005) デモセッション     1620 - 1621   2005.07  [Refereed]

識別器選択のための入力空間分割法に関する検討

林拓, 中村恭之, 和田俊和

画像の認識・理解シンポジウム(MIRU2005)     859 - 866   2005.07  [Refereed]

計画と実行の反復による車輪型移動ロボットの滑り運動学習

川原輝美, 中村恭之, 和田俊和

第10回ロボティクスシンポジア予稿集     405 - 409   2005.03  [Refereed]

MDL-based nonlinear regression tree

UEZU Yoshimi, NAKAMURA Takayuki, WADA Toshikazu

IEICE technical report. Natural language understanding and models of communication ( 社団法人電子情報通信学会 )  104 ( 667 ) 13 - 18   2005.02

　View Summary

In the previous work, we proposed the "PaLM-Tree" which is a kind of linear regression tree. It improves some drawbacks which the conventional linear regression tree method have. When it is applied for estimating a complex nonlinear mapping, it splits the input space into too much regions. In such case, the generalization performance of the PaLM-tree tends to be worse. In order to improve the generalization performance of the PaLM-tree, we proposed a MDL-based nonlinear regression tree. Instead of using linear regression functions, this method uses polynomial functions. Furthermore, it util...

Space Partitioning Method for Classifier Selection

HAYASHI Hiromu, NAKAMURA Takayuki, WADA Toshikazu

IEICE technical report. Natural language understanding and models of communication ( 社団法人電子情報通信学会 )  104 ( 667 ) 7 - 12   2005.02

　View Summary

Ensemble learing is a method for constructing a strong crassifier from multiple weak classifiers. In this method, the accuracy of the strong classifier is improved, however, the classification speed and memory consumption degenerate. This is because that the strong classifier requires outputs of all weak classifiers for a single input. Our approach consists of two ideas : decomposing the input space into boxes and selecting the best classifier for each box from a certain classifier set. In this approach, since a single classifier is selected for a single input, the classification speed is a...

1P2-S-067 High performance control of active camera head using PaLM-tree(Evolution and Learning for Robotics 3,Mega-Integration in Robotics and Mechatronics to Assist Our Daily Lives)

Sakata Yoshio, Nakamura Takayuki, Wu Haiyuan, Wada Toshikazu

The Proceedings of JSME annual Conference on Robotics and Mechatronics (Robomec) ( The Japan Society of Mechanical Engineers )  2005 ( 0 ) 133 - 133   2005

DOI

Regression Trees and Its Application to Classification Problems

Wada Toshikazu, Nakamura Takayuki

IPSJ SIG Notes. CVIM ( 一般社団法人情報処理学会 )  2004 ( 91 ) 203 - 210   2004.09

　View Summary

This paper presents evolutional history of regression trees, a novel regression tree called PaLM-tree, and its application to Computer Vision and Pattern Recognition problems. Regression tree is a tree representation of nonlinear function, which is mainly investigated in the fields of Machine Learning and Data Mining. The original regression tree is a simple tree representation holding function values at leaves (terminated nodes). Linear regression tree extends the function values to linear regression coefficients. Since these regression trees are designed for rule-extraction, most of them ...

Regression Trees and Its Application to Classification Problems

Wada Toshikazu, Nakamura Takayuki

Technical report of IEICE. PRMU ( 社団法人電子情報通信学会 )  104 ( 291 ) 33 - 40   2004.09

　View Summary

This paper presents evolutional history of regression trees, a novel regression tree called PaLM-tree, and its application to Computer Vision and Pattern Recognition problems. Regression tree is a tree representation of nonlinear function, which is mainly investigated in the fields of Machine Learning and Data Mining. The original regression tree is a simple tree representation holding function values at leaves (terminated nodes). Linear regression tree extends the function values to linear regression coefficients. Since these regression trees are designed for rule-extraction, most of them ...

回帰木を用いた非線形写像の学習と識別問題への応用

和田,中村

電子情報通信学会PRMU研究会資料 ( Information Processing Society of Japan (IPSJ) )  2004 ( 91 ) 203 - 210   2004.09  [Refereed]

　View Summary

This paper presents evolutional history of regression trees, a novel regression tree called PaLM-tree, and its application to Computer Vision and Pattern Recognition problems. Regression tree is a tree representation of nonlinear function, which is mainly investigated in the fields of Machine Learning and Data Mining. The original regression tree is a simple tree representation holding function values at leaves (terminated nodes). Linear regression tree extends the function values to linear regression coefficients. Since these regression trees are designed for rule-extraction, most of them are single valued for representing nonlinear scalar functions. PaLM-tree is a multi-valued extension of the linear regression tree with Split-and-Merge tree construction. PaLM-tree can be applied to varieties of nonlinear mapping learning problems. This paper also presents applications of PaLM-tree involving image segmentation, camera calibration and imitating classifiers.

I-067 Face Direction Estimation using Nonlinear Map Learning Algorithm "PaLM-tree"

Satoh Tetsu, Wada Toshikazu, Nakamura Takayuki

情報科学技術フォーラム一般講演論文集 ( Forum on Information Technology )  3 ( 3 ) 157 - 158   2004.08

カメラ同期・対応点の必要のない複数ステレオカメラのキャリブレーション法

飯塚 健男, 中村 恭之, 和田 俊和

知能メカトロニクスワークショップ講演論文集 ( 〔精密工学会〕 )  9   157 - 162   2004.08