跳到主要內容

臺灣博碩士論文加值系統

(18.97.9.172) 您好!臺灣時間:2025/02/14 02:39
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果 :::

詳目顯示

我願授權國圖
: 
twitterline
研究生:陳煥元
研究生(外文):Huan-Yuan Chen
論文名稱:於預測時主動式獲取不完全資料之資料特徵值
論文名稱(外文):Prediction-time Active Feature-value Acquisition GivenIncomplete Data
指導教授:林守德林守德引用關係
指導教授(外文):Shou-De Lin
口試日期:2017-06-26
學位類別:碩士
校院名稱:國立臺灣大學
系所名稱:資訊工程學研究所
學門:工程學門
學類:電資工程學類
論文種類:學術論文
論文出版年:2017
畢業學年度:105
語文別:英文
論文頁數:24
中文關鍵詞:於預測時主動式獲取不完全資料之資料特徵值智能採集策略性能預測
外文關鍵詞:prediction-time active feature-value acquisitionintelligent acquisition strategyprediction performance
相關次數:
  • 被引用被引用:0
  • 點閱點閱:182
  • 評分評分:
  • 下載下載:0
  • 收藏至我的研究室書目清單書目收藏:0
於預測時主動式獲取不完全資料之資料特徵值可為在商業實務上一個重要的應用。P. Melville 於2008年提出如何於預測時主動獲取含不完全資料的資料點之特徵值。然而,之前的相關研究並沒有專注於研究獲取單一資料點的缺失特徵值之方法,而是研究如何獲取完整的單一資料點。在這篇碩士論文中, 我們展現一個由已訓練之分類器獲取單一測試資料點之特徵值架構。首先,計算已訓練分類器對已填值之資料的影響力。然後,經填值模組一階泰勒展開式估計從填值域的此影響力線性變換至使用者輸入域的影響力。最後,根據此轉換的影響力構造獲取特徵值之問題,以向使用者詢問額外的資訊。我們的實驗結果顯示使用此詢問架構與隨機詢問相比使分類器效能顯著地提升。
Prediction-time feature-value acquisition is an important application in business customer service division. The original framework provided by P. Melville illustrates a strategy for requesting feature value for missing feature-value instances. Nevertheless, no previous work focuses on the method of acquiring the value of one feature other than the whole instance. In this thesis, we demonstrate a framework for querying missing feature for a single test instance utilizing trained classifier. First, calculate the influence provided by trained classifier over imputed input. Then, transform the influence in imputed domain to input domain through first-order Taylor series approximation of imputation method. By this converted influence, feature acquisition questions can be formed and send to user for additional information. Our experiment results show that after this querying strategy, the performance is significantly improved compare to random inquiry.
口試委員會審定書i
誌謝ii
摘要iii
Abstract iv
1 Introduction 1
1.1 Introduction to active feature-value acquisition . . . . . . . . . . . . . . 1
1.2 Introduction to feature imputation . . . . . . . . . . . . . . . . . . . . . 2
1.3 Thesis organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 Related Works 4
3 Methodology 7
3.1 proposed framework for data imputation and influence computation . . . 7
3.2 incomplete data instance imputation . . . . . . . . . . . . . . . . . . . . 9
3.3 Influence transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.4 General framework for prediction-time AFA given incomplete data . . . . 14
4 Experiment Results 15
4.1 Experimental setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
4.1.1 Data description and implementation . . . . . . . . . . . . . . . 15
4.1.2 Model setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
4.1.3 Evaluation metric . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.2 Experiment and results . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.2.1 Comparison with competitors . . . . . . . . . . . . . . . . . . . 17
5 Conclusion 20
A Supplementary Materials 21
Bibliography 23
[1] M. Abdella and T. Marwala. Treatment of missing data using neural networks and genetic algorithms. In Neural Networks, 2005. IJCNN’05. Proceedings. 2005 IEEE International Joint Conference on, volume 1, pages 598–603. IEEE, 2005.
[2] P. D. Allison. Multiple imputation for missing data: A cautionary tale. Sociological methods & research, 28(3):301–309, 2000.
[3] M. H. Hassoun. Fundamentals of artificial neural networks. MIT press, 1995.
[4] S. Ji and L. Carin. Cost-sensitive feature acquisition and classification. Pattern Recognition, 40(5):1474–1485, 2007.
[5] P. Kanani and P. Melville. Prediction-time active feature-value acquisition for costeffective customer targeting. Advances In Neural Information Processing Systems(NIPS), 2008.
[6] M. Lichman. UCI machine learning repository, 2013.
[7] R. J. Little and D. B. Rubin. Statistical analysis with missing data. John Wiley & Sons, 2014.
[8] P. Melville, F. Provost, M. Saar-Tsechansky, and R. Mooney. Economical active feature-value acquisition through expected utility estimation. In Proceedings of the 1st international workshop on Utility-based data mining, pages 10–16. ACM, 2005.
[9] P. Melville, M. Saar-Tsechansky, F. Provost, and R. Mooney. Active feature-value acquisition for classifier induction. In Data Mining, 2004. ICDM’04. Fourth IEEE International Conference on, pages 483–486. IEEE, 2004.
[10] D. Mladenić, J. Brank, M. Grobelnik, and N. Milic-Frayling. Feature selection using linear classifier weights: interaction with classification models. In Proceedings of the 27th annual international ACM SIGIR conference on Research and development in information retrieval, pages 234–241. ACM, 2004.
[11] D. B. Rubin. Multiple imputations in sample surveys-a phenomenological bayesian approach to nonresponse. In Proceedings of the survey research methods section of the American Statistical Association, volume 1, pages 20–34. American Statistical Association, 1978.
[12] M. Saar-Tsechansky, P. Melville, and F. Provost. Active feature-value acquisition. Management Science, 55(4):664–684, 2009.
[13] P. Vincent, H. Larochelle, Y. Bengio, and P.-A. Manzagol. Extracting and composing robust features with denoising autoencoders. In Proceedings of the 25th international conference on Machine learning, pages 1096–1103. ACM, 2008.
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top