English  |  正體中文  |  简体中文  |  全文筆數/總筆數 : 80990/80990 (100%)
造訪人次 : 41262204      線上人數 : 137
RC Version 7.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
搜尋範圍 查詢小技巧:
  • 您可在西文檢索詞彙前後加上"雙引號",以獲取較精準的檢索結果
  • 若欲以作者姓名搜尋,建議至進階搜尋限定作者欄位,可獲得較完整資料
  • 進階搜尋


    請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/78653


    題名: 以Jacobian Matrix為基礎之深度學習機器之研發及其於姿態動作之辨識應用;The Development of a Deep Jacobian-Matrix-Based Learning Machine and Its Applications in Posture Recognition
    作者: 蘇木春
    貢獻者: 國立中央大學資訊工程系
    關鍵詞: 姿態辨識系統;學習機器;類神經網路;活動度量測;posture recognition system;learning machine;neural network;measurements of the activities
    日期: 2018-12-19
    上傳時間: 2018-12-20 13:42:03 (UTC+8)
    出版者: 科技部
    摘要: 此計畫是一個三年期的計畫,整體的總目標是要開發可應用於不同情境之姿態辨識系統。第一年的計畫目標是研發一種簡單而有效的解決方案來解決類神經網路或學習機器(learning machine)的參數學習(即每層中的鍵結值權重)和結構學習(即網路結構)的問題。由於具備許多吸引人的特性,類神經網路成為解決不同類型問題的一個很有效的工具。在很大程度上,類神經網路的性能取決於它們是否能夠為參數學習和結構學習的問題提供有吸引力的解決方案。這兩種學習可以同時或分開進行。在本計畫中的第一年,我們希望研發一種以Jacobian Matrix為基礎之深度學習機器 (a Deep Jacobian Matrix-based Learning Machine,簡稱Deep JMLM) 為上述兩種學習提供了一個有吸引力的解決方案。 Deep JMLM的網路結構可以遞增式地增加網路節點,並且提出基於Jacobian矩陣的學習方法來有效地估計相應的網路參數。接下來的兩年就是要以此Deep JMLM為基礎來實現姿態辨識系統。  近年來拜醫療科技的進步,台灣社會面臨人口結構高齡化的問題,獨居老人照護問題也比以往更加需要被關注,如何有效且即時地對獨居老人的活動量進行客觀的量測是目前很重要的一個議題。因此,第二年要針對獨居老人開發一套居家生活動作辨識系統,預計要能辨識十種不同之居家生活動作,最後系統可以提供一個可供分析的活動量評估報告。而第三年的計畫目標對象則是針對學童開發一套學習姿態辨識系統,可以全程掌握孩童念書或上課時的專注的程度,並能適時提醒孩童眼睛是否距離書本太近或是有不良之閱讀坐姿(如:駝背、手撐頭等),其目標是要減輕父母親的負擔和掌握孩童學習的狀態。 ;This is a three-year project. The main goal of this project is to develop posture recognition systems which can be applied in different scenarios. In the first year, we aim to develop a workable solution to the two kinds of learning problems involved in training a neural network: the parameter learning and structure learning problems. Owing to many appealing properties, neural networks provide a natural basis for solving different kinds of problems. The performance of neural networks greatly depends on whether they can provide appealing solutions to the problems of the parameter learning (i.e., the connecting weights in each layer) and the structure learning (i.e., the network structure). These two kinds of learning can be performed simultaneously or separately. In this project, we will develop a Deep Jacobian Matrix-based Learning Machine (Deep JMLM) to provide an appealing solution to the aforementioned two kinds of learning. The network structure of a JMLM can be incrementally constructed and a Jacobian-matrix-based learning method is proposed to efficiently estimate the corresponding network parameters. We then will apply the Deep JMLM to build posture recognition systems in the following two years.  In recent years, due to advances in medical science and technology, Taiwan society is facing the problem of aging population structure. The issue of the elderly care for the elderly living alone needs more attention than ever before. How to conduct objective and effective measurements of the activities of the elderly living alone is a very important issue nowadays. Therefore, the main goal of the second year is to develop a posture recognition system for the elderly living alone. Ten different home-based activities are expected to be identified. Finally, an evaluation report for analysis will be provided. The goal of the third year is to develop a reading posture recognition system for children. The system can fully monitor the degree of concentration of children in their studies and promptly remind them whether the eyes are too close to the books or they have poor reading postures (e.g., humpback, one hand supporting his/her head). The goal of this system is to reduce the burden on parents and master the status of children learning.
    關聯: 財團法人國家實驗研究院科技政策研究與資訊中心
    顯示於類別:[資訊工程學系] 研究計畫

    文件中的檔案:

    檔案 描述 大小格式瀏覽次數
    index.html0KbHTML259檢視/開啟


    在NCUIR中所有的資料項目都受到原著作權保護.

    社群 sharing

    ::: Copyright National Central University. | 國立中央大學圖書館版權所有 | 收藏本站 | 設為首頁 | 最佳瀏覽畫面: 1024*768 | 建站日期:8-24-2009 :::
    DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - 隱私權政策聲明