博碩士論文 etd-0929121-131855 詳細資訊


[回到前頁查詢結果 | 重新搜尋]

姓名 王珮芬(Pei-Fen Wang) 電子郵件信箱 E-mail 資料不公開
畢業系所 電子商務與商業分析數位學習碩士在職專班(Online Master of Business Administration in Electronic Commerce and Business Analytics)
畢業學位 碩士(Master) 畢業時期 109學年第2學期
論文名稱(中) 基於可解釋深度神經網路的流程模型
論文名稱(英) Interpretable Process Modeling based on deep neural networks
檔案
  • etd-0929121-131855.pdf
  • 本電子全文僅授權使用者為學術研究之目的,進行個人非營利性質之檢索、閱讀、列印。
    請遵守中華民國著作權法之相關規定,切勿任意重製、散佈、改作、轉貼、播送,以免觸法。
    論文使用權限

    紙本論文:立即公開

    電子論文:校內校外完全公開

    論文語文/頁數 中文/53
    統計 本論文已被瀏覽 265 次,被下載 51 次
    摘要(中) 隨著環境與科技的快速發展,促使公司不斷進行跨企業整合併購、外包與數位創新,也期望能為公司與關係企業開創其獨有的生態圈並提升其競爭力。不過,這樣的整併與外包也常使企業及其經營團隊不易梳理、掌握企業整體營運過程並承受著較高的營運風險與管理成本。
    由於企業營運有效性評估並不只限於其結果的數據分析,更重要的是集團營運過程的順暢度與績效。故企業的營運決策亦應考量「流程」的角度,對營運執行過程進行監控與管理,以提升企業營運績效並控制營運成本。
    然企業體內部業務分立,作業流程如千經萬緯的網路,需仰賴資訊系統處理各類作業,而企業軟體系統設計保留各類流程資料,並存儲於事件日誌檔(event log file)作為日後稽查所需。在歐美國家例如IBM,早在多年前即已投入事件日誌的數據分析與研究,以協助尋找流程瓶頸或異常偵測。
    本研究將透過系統事件日誌(event log)建立流程模型 (process model),並藉流程模型萃取事件流程序列(sequence)與其執行結果,透過機器學習中的邏輯斯回歸 (Logistic Regression)、決策樹 (Decision Tree)、隨機森林 (Random Forest) 及深度神經網路中的長期短期記憶 (Long Short-Term memory, LSTM) 及一維卷積神經網路 (1 Dimensional Convolutional Neural Network, 1D CNN) 等演算法創建預測模型並進行模型評量。
    最後,將透過可視化模型解釋器LIME (Local Interpretable Model-Agnostic Explanations) 對訓練完成的深度神經網路預測模型進行解釋,以期提供管理階層可信賴的案件執行預判結果。讓企業在面對具營運風險且須反覆執行的行動案件時,於其執行時儘早獲得警示或啟動預防措施,將有利於提供事件後續的適處並預留因應作業的空間。
    摘要(英) With the rapid development of environment and technology, companies continue to carry out cross-enterprise integration mergers and acquisitions, outsourcing and digital innovation, and hope to create their own unique ecosystem and enhance their competitiveness for companies and related companies. However, this also makes it difficult for the company and its management team to sort out, master the overall operation process, and bear higher operational risks and management costs.
    Because the evaluation of the effectiveness of a company's operations is not limited to the data analysis of its results, it is more important to the smoothness and performance of the group's operations. Therefore, the company's operational decision-making should also consider the perspective of "process", monitor and manage the operational execution process, so as to improve the company's operational performance and control operating costs.
    This research will establish a process model through the event log, and use it to extract the sequence of events and their execution results and use Long Short-Term Memory (LSTM) in the deep neural network and One Dimensional Convolutional Neural Network (1D CNN) process model establishment and model evaluation the interpretable model LIME (Local Interpretable Model-Agnostic Explanations) is used to explain the trained process model in order to provide a predictive model that the management can trust.
    關鍵字(中)
  • 解釋模型
  • 神經網路
  • 流程模型
  • 流程序列
  • RNN
  • 1D CNN
  • LIME
  • 關鍵字(英)
  • Process Modeling
  • Sequence Analysis
  • RNN
  • 1D CNN
  • LIME
  • 論文目次 論文審定書 i
    誌 謝 ii
    摘 要 iii
    Abstract iv
    目錄 v
    圖目錄 vii
    表目錄 ix
    第 一 章 緒論 1
    1.1 研究背景 1
    1.2 研究動機 2
    1.3 研究目的 2
    第 二 章 文獻探討 3
    2.1 流程模型 (Process Model) 3
    2.2 文本模型 (Text Model) 5
    2.3 機器學習 (Machine Learning) 6
    2.4 深度學習 (Deep Learning) 8
    2.5 局部解釋模型 10
    第 三 章 研究方法與步驟 13
    3.1 研究流程 13
    3.2 研究資料 14
    3.3 研究方法 18
    3.3 研究架構 24
    第 四 章 研究結果與分析 26
    4.1 創建預測模型 26
    4.2 模型評估 28
    4.3 模型解釋 30
    4.4 研究討論與限制 37
    第 五 章 結論與未來發展趨勢 39
    參考文獻 40
    附表 42
    參考文獻 [1]  W. van der Aalst, Process Mining, Data Science in Action, 2nd ed. Berlin Heidelberg: Springer-Verlag, 2016. doi: 10.1007/978-3-662-49851-4.
    [2] ‘Deep Learning with R’, Manning Publications. https://www.manning.com/books/deep-learning-with-r (accessed Jun. 11, 2021).
    [3]  F. A. Gers, J. Schmidhuber, and F. Cummins, ‘Learning to Forget: Continual Prediction with LSTM’, Neural Comput., vol. 12, no. 10, pp. 2451–2471, Oct. 2000, doi: 10.1162/089976600300015015.
    [4]  H. Lee and J. Song, ‘Introduction to convolutional neural network using Keras; an understanding from a statistician’, Commun. Stat. Appl. Methods, vol. 26, pp. 591–610, Nov. 2019, doi: 10.29220/CSAM.2019.26.6.591.
    [5]  M. T. Ribeiro, S. Singh, and C. Guestrin, ‘“Why Should I Trust You?”: Explaining the Predictions of Any Classifier’, ArXiv160204938 Cs Stat, Aug. 2016, Accessed: Jun. 12, 2021. [Online]. Available: http://arxiv.org/abs/1602.04938
    [6]  F. Almeida and G. Xexéo, ‘Word Embeddings: A Survey’, ArXiv190109069 Cs Stat, Jan. 2019, Accessed: Jul. 11, 2021. [Online]. Available: http://arxiv.org/abs/1901.09069
    [7]  B. van Dongen and F. (Florian) Borchert, ‘BPI Challenge 2018’, 2018, doi: 10.4121/uuid:3301445f-95e8-4ff0-98a4-901f1f204972.
    [8] ‘R: a language and environment for statistical computing’, Feb. 10, 2015. https://www.gbif.org/zh-tw/tool/81287/r-a-language-and-environment-for-statistical-computing (accessed Jun. 16, 2021).
    [9]  Janssenswillen, G., Depaire, B., Swennen, M, .Jans, M., and Vanhoof, K., ‘bupaR: Enabling reproducible business process analysis.’, Business Process Analysis_bupaR, 2019. http://www.bupar.net/index.html (accessed Jun. 16, 2021).
    [10] D. Breuker, P. Delfmann, M. Matzner, and J. Becker, ‘Designing and Evaluating an Interpretable Predictive Modeling Technique for Business Processes’, 2015, Accessed: Jun. 13, 2021. [Online]. Available: https://www.wi.uni-muenster.de/de/forschung/publikationen/97048
    [11] S. Gholizadeh and N. Zhou, ‘Model Explainability in Deep Learning Based Natural Language Processing’, ArXiv210607410 Cs, Jun. 2021, Accessed: Jun. 26, 2021. [Online]. Available: http://arxiv.org/abs/2106.07410
    口試委員
  • 胡雅涵 - 召集委員
  • 林耕霈 - 委員
  • 康藝晃 - 指導教授
  • 口試日期 2021-07-23 繳交日期 2021-10-29

    [回到前頁查詢結果 | 重新搜尋]


    如有任何問題請與論文審查小組聯繫