Powerful Strategy for Point Process Data Analysis and the Archive of Models

用于点过程数据分析和模型存档的强大策略

• 批准号: 15300095
• 负责人: SHIBATA Ritei
• 金额: $ 6.59万
• 依托单位: Keio University
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for Scientific Research (B)
• 财政年份: 2003
• 资助国家: 日本
• 起止时间: 2003 至 2006
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-15300095/
• 关键词: Point Process; Modelling; Data; Textile Plot; Multiplicative Correlation; Hidden Markov Model; Neural Network; Maximum Likelihood Solution; 隠れ点過程; 時系列; 神経細胞; 外国為替; 最尤法; クラスタ; 両側指数分布

Various Exploratory Data Analysis techniques have been developed particularly for time point data. The aim of this research is to provide a general way of building appropriate models for the given data. The models we have investigated are clustering marked point process, hidden Markov point process, hidden semi Markov point process, mixed point process, multiple point process and zero reset intensity model. We have developed not only mathematical theory but also algorithms of solving the maximum likelihood equation and a total environment for supporting the process of model building. The targeted phenomena are neural activation, hear beat, rain fall, earthquakes, customer visit to a beauty parlor, new bid price of foreign exchange. The environment is based on DandD and one of the features is a new data visualization technique called Textile plot.

各种探索性数据分析技术已经开发出来，特别是针对时间点数据。本研究的目的是为给定的数据提供一种建立适当模型的通用方法。我们研究的模型有聚类标记点过程、隐马尔可夫点过程、隐半马尔可夫点过程、混合点过程、多点过程和零重置强度模型。我们不仅开发了数学理论，而且还开发了求解最大似然方程的算法和支持模型构建过程的总体环境。目标现象是神经激活、心跳、下雨、地震、顾客去美容院、外汇新出价。该环境是基于ddd的，其特点之一是一种新的数据可视化技术——纺织图。

项目成果

Effectiveness of Stochastic Neural Network for Prediction of Fall or Rise of TOPIX

随机神经网络预测东证指数涨跌的有效性

• 发表时间: 2005
• 期刊: Asian-Pacific Financial Markets 15
• 作者: S.Kamitsuji;R.Shibata
• 通讯作者: R.Shibata

Miki MIYAZAWA, Ritel SHIBATA: "Statistical Aspects of Instantaneous FX Bid Prices"The 3^<rd> International Conference on Frinancial Engineering and Statistical Finance. 3. 66-73 (2003)

Miki MIYAZAWA、Ritel Shibata：“瞬时外汇买入价的统计方面”第三届金融工程和统计金融国际会议。

島津秀康, 柴田里程: "野鳥羽数時系列の局所回帰による分解とその環境要因変化との関連性"2003年度統計関連学会連合大会講演報告集. 3. 381-382 (2003)

Hideyasu Shimazu、Rijo Shibata：“通过局部回归分解野生鸟类种群时间序列及其与环境因素变化的关系”2003 年统计学会联合会会议记录 3. 381-382 (2003)。

Efficient Learning Algorithm Based on Maximum Likelihood Principle for 1VIultilayer Feedforward Stochastic Neural Network

基于最大似然原理的1VI多层前馈随机神经网络高效学习算法

• 发表时间: 2007
• 期刊: Neural Processing Letters (to appear)
• 作者: S.Kamitsuji;R.Shibata
• 通讯作者: R.Shibata

High Dimensional Data Visualization : the Textile Plot

高维数据可视化：纺织图

• 发表时间: 2006
• 期刊: Computational Statistics & Data Analysis (submitted)(未定)
• 作者: N.Kumasaka;R.Shibata
• 通讯作者: R.Shibata