Loading...
|
Please use this identifier to cite or link to this item:
https://nccur.lib.nccu.edu.tw/handle/140.119/153147
|
| Title: | 在時間序列迴歸中應用漂移檢測器
Empirical Drift Detector for Time Series Regression Tasks |
| Authors: | 莊宗縉
Chuang, Tsung-Chin |
| Contributors: | 林怡伶
Lin, Yi-Ling
莊宗縉
Chuang, Tsung-Chin |
| Keywords: | 概念漂移
漂移檢測
時間序列預測
機器學習
Concept drift
Drift detection
Time series forecasting
Machine Learning |
| Date: | 2024 |
| Issue Date: | 2024-09-04 14:03:07 (UTC+8) |
| Abstract: | 在本論文中，我們介紹了經驗漂移檢測器（EDD），旨在解決迴歸任務中概念漂移檢測的不足之處。EDD採用了三西格馬法則（three-sigma rule）來分類預測誤差，並識別超出兩個標準差區間的異常值。前期測試已證實我們的機器學習預測模型誤差符合常態分布。EDD使用一個固定長度的滑動視窗，並結合時間衰減函數來計算時間加權得分，當該分數超過預先設定的閾值時，即表明可能發生了概念漂移。為了應對重複性漂移，EDD會將模型儲存至模型庫，並在檢測到概念漂移時評估過往模型的性能，適時啟用舊模型來適應當前概念，以省去模型更新的時間，進而提升計算效率。我們的主要假設是，將EDD與極限學習機（ELM）結合，可以實現與在線序列極限學習機（OS-ELM）相當的準確性，同時降低計算成本。實驗證據支持了這個假設，顯示在具有重複性漂移的資料集和真實世界資料集中，ELM+EDD在準確性上表現出色且計算需求低。我們的次要假設是，EDD在識別急遽型漂移方面優於傳統的漂移檢測方法，且誤報率較低。與漂移檢測方法（DDM）和早期漂移檢測方法（EDDM）的比較分析表明，ELM+EDD在平均漂移檢測距離和誤報率方面具有更優異的表現。這些研究結果證明EDD是一種實用且效率高的迴歸任務漂移檢測方法。
In this paper, we introduce the Empirical Drift Detector (EDD) to address deficiencies in drift detection for regression tasks. The EDD employs the three-sigma rule to classify prediction errors, identifying anomalies beyond the two-sigma interval. A pretest confirmed the normal distribution of these errors. Utilizing a sliding window with a fixed length as a forgetting mechanism, EDD integrates a time decay function to compute a time-weighted score, indicating potential concept drift when a predefined threshold is surpassed.
To manage recurring drifts, EDD archives models in a repository and evaluates their performance upon drift detection, thereby optimizing computational efficiency. Our primary hypothesis posits that integrating EDD with the Extreme Learning Machine (ELM) can achieve the accuracy of the Online Sequential Extreme Learning Machine (OS-ELM) while reducing computational costs. Empirical evidence supports this hypothesis, demonstrating that ELM+EDD attains comparable accuracy with significantly lower computational demands, particularly in datasets with recurrent drift and real-world applications such as the gold price dataset.
Our secondary hypothesis asserts that EDD outperforms conventional drift detection methods in identifying abrupt drifts with a reduced frequency of false alarms. Comparative analyses with the Drift Detection Method (DDM) and Early Drift Detection Method (EDDM) reveal that ELM+EDD exhibits superior performance in terms of average distance to drift detection and false alarm rates. These findings establish EDD as an effective and efficient drift detection method for regression tasks. |
| Reference: | Agrahari, S., & Singh, A. K. (2022). Concept Drift Detection in Data Stream Mining: A literature review. In Journal of King Saud University - Computer and Information Sciences (Vol. 34, Issue 10, pp. 9523–9540). King Saud bin Abdulaziz University. https://doi.org/10.1016/j.jksuci.2021.11.006
Alzahrani, S., Alashri, S., Koppela, A. R., Davulcu, H., & Toroslu, I. (2015). A network-based model for predicting hashtag breakouts in Twitter. Social Computing, Behavioral-Cultural Modeling, and Prediction: 8th International Conference, SBP 2015, Washington, DC, USA, March 31-April 3, 2015. Proceedings 8, 3–12.
Baena-Garcıa, M., del Campo-Ávila, J., Fidalgo, R., Bifet, A., Gavalda, R., & Morales-Bueno, R. (2006). Early drift detection method. Fourth International Workshop on Knowledge Discovery from Data Streams, 6, 77–86.
Baier, L., Hofmann, M., Kühl, N., Mohr, M., & Satzger, G. (2020). Handling Concept Drifts in Regression Problems--the Error Intersection Approach. ArXiv Preprint ArXiv:2004.00438.
Bakirov, R., Gabrys, B., & Fay, D. (2015). On sequences of different adaptive mechanisms in non-stationary regression problems. 2015 International Joint Conference on Neural Networks (IJCNN), 1–8.
Bakker, J. (2012). Handling abrupt changes in evolving time-series data. https://doi.org/10.6100/IR740215
Basseville, M., & Nikiforov, I. V. (1993). Detection of abrupt changes: theory and application (Vol. 104). prentice Hall Englewood Cliffs.
Bayram, F., Ahmed, B. S., & Kassler, A. (2022). From concept drift to model degradation: An overview on performance-aware drift detectors. Knowledge-Based Systems, 245, 108632.
Cabello-López, T., Cañizares-Juan, M., Carranza-García, M., Garcia-Gutiérrez, J., & Riquelme, J. C. (2022). Concept Drift Detection to Improve Time Series Forecasting of Wind Energy Generation. In P. García Bringas, H. Pérez García, F. J. Martínez de Pisón, J. R. Villar Flecha, A. Troncoso Lora, E. A. de la Cal, Á. Herrero, F. Martínez Álvarez, G. Psaila, H. Quintián, & E. Corchado (Eds.), Hybrid Artificial Intelligent Systems (pp. 133–140). Springer International Publishing.
Cavalcante, R. C., & Oliveira, A. L. I. (2015). An approach to handle concept drift in financial time series based on Extreme Learning Machines and explicit Drift Detection. 2015 International Joint Conference on Neural Networks (IJCNN), 1–8.
Chen, Z., Han, M., Wu, H., Li, M., & Zhang, X. (2023). A multi-level weighted concept drift detection method. The Journal of Supercomputing, 79(5), 5154–5180.
Choudhary, M., Jain, S., & Arya, G. (2022). Classical Models vs Deep Leaning: Time Series Analysis. International Conference on Advancements in Interdisciplinary Research, 276–286.
Cohen, E., & Strauss, M. (2003). Maintaining time-decaying stream aggregates. Proceedings of the Twenty-Second ACM SIGMOD-SIGACT-SIGART Symposium on Principles of Database Systems, 223–233.
Demšar, J., & Bosnić, Z. (2018). Detecting concept drift in data streams using model explanation. Expert Systems with Applications, 92, 546–559.
Ding, S., Zhao, H., Zhang, Y., Xu, X., & Nie, R. (2015). Extreme learning machine: algorithm, theory and applications. Artificial Intelligence Review, 44, 103–115.
Gama, J., Medas, P., Castillo, G., & Rodrigues, P. (2004). Learning with drift detection. Advances in Artificial Intelligence–SBIA 2004: 17th Brazilian Symposium on Artificial Intelligence, Sao Luis, Maranhao, Brazil, September 29-Ocotber 1, 2004. Proceedings 17, 286–295.
Gama, J., Zliobaite, I., Bifet, A., Pechenizkiy, M., & Bouchachia, A. (2014). A survey on concept drift adaptation. In ACM Computing Surveys (Vol. 46, Issue 4). Association for Computing Machinery. https://doi.org/10.1145/2523813
Gonçalves, P. M., De Carvalho Santos, S. G. T., Barros, R. S. M., & Vieira, D. C. L. (2014). A comparative study on concept drift detectors. In Expert Systems with Applications (Vol. 41, Issue 18, pp. 8144–8156). https://doi.org/10.1016/j.eswa.2014.07.019
Gözüaçık, Ö., & Can, F. (2021). Concept learning using one-class classifiers for implicit drift detection in evolving data streams. Artificial Intelligence Review, 54(5), 3725–3747. https://doi.org/10.1007/s10462-020-09939-x
Harries, M., & Horn, K. (1995). Detecting concept drift in financial time series prediction using symbolic machine learning. AI-CONFERENCE-, 91–98.
Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural Computation, 9(8), 1735–1780.
Hu, Y., Feng, B., Zhang, X., Ngai, E. W. T., & Liu, M. (2015). Stock trading rule discovery with an evolutionary trend following model. Expert Systems with Applications, 42(1), 212–222.
Huang, G.-B., Zhou, H., Ding, X., & Zhang, R. (2011). Extreme learning machine for regression and multiclass classification. IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), 42(2), 513–529.
Huang, G.-B., Zhu, Q.-Y., & Siew, C.-K. (2006). Extreme learning machine: theory and applications. Neurocomputing, 70(1–3), 489–501.
Hyndman, R., Koehler, A. B., Ord, J. K., & Snyder, R. D. (2008). Forecasting with exponential smoothing: the state space approach. Springer Science & Business Media.
Jagait, R. K., Fekri, M. N., Grolinger, K., & Mir, S. (2021). Load Forecasting Under Concept Drift: Online Ensemble Learning With Recurrent Neural Network and ARIMA. IEEE Access, 9, 98992–99008. https://doi.org/10.1109/ACCESS.2021.3095420
Krawczyk, B., & Cano, A. (2018). Online ensemble learning with abstaining classifiers for drifting and noisy data streams. Applied Soft Computing, 68, 677–692.
Krishna, V. R., & Boddu, S. B. (2023). Financial Fraud Detection using Improved Artificial Humming Bird Algorithm with Modified Extreme Learning Machine. International Journal on Recent and Innovation Trends in Computing and Communication. https://api.semanticscholar.org/CorpusID:259850203
Liang, N.-Y., Huang, G.-B., Saratchandran, P., & Sundararajan, N. (2006). A fast and accurate online sequential learning algorithm for feedforward networks. IEEE Transactions on Neural Networks, 17(6), 1411–1423.
Lima, M., Neto, M., Filho, T. S., & Roberta, R. A. (2022). Learning Under Concept Drift for Regression - A Systematic Literature Review. IEEE Access, 10, 45410–45429. https://doi.org/10.1109/ACCESS.2022.3169785
Liu, Z., Loo, C. K., & Seera, M. (2019). Meta-cognitive Recurrent Recursive Kernel OS-ELM for concept drift handling. Applied Soft Computing Journal, 75, 494–507. https://doi.org/10.1016/j.asoc.2018.11.006
Lu, J., Liu, A., Dong, F., Gu, F., Gama, J., & Zhang, G. (2018). Learning under concept drift: A review. IEEE Transactions on Knowledge and Data Engineering, 31(12), 2346–2363.
Lu, N., Zhang, G., & Lu, J. (2014). Concept drift detection via competence models. Artificial Intelligence, 209, 11–28.
Luengo, J., Sanchez-Tarrago, D., Prati, R. C., & Herrera, F. (2021). Multiple instance classification: Bag noise filtering for negative instance noise cleaning. Information Sciences, 579, 388–400.
Mayaki, M. Z. A., & Riveill, M. (2022). Autoregressive based drift detection method. 2022 International Joint Conference on Neural Networks (IJCNN), 1–8.
Mirza, B., Lin, Z., & Liu, N. (2015). Ensemble of subset online sequential extreme learning machine for class imbalance and concept drift. Neurocomputing, 149, 316–329.
Mishra, J. P., & Das, S. R. (2022). Gold Price Prediction Using an Evolutionary Extreme Learning Machine. In Advances in Data Science and Management: Proceedings of ICDSM 2021 (pp. 93–99). Springer.
Museba, T., Nelwamodo, F., & Ouhada, K. (2019). An Adaptive Ensemble Classifier for Handling Recurring Concepts. 2019 International Multidisciplinary Information Technology and Engineering Conference (IMITEC), 1–7.
Nguyen, K.-T., Tran, T., Nguyen, A.-D., Phan, X.-H., & Ha, Q.-T. (2023). Concept Drift Detection in Data Stream: Ensemble Learning Method for Detecting Gradual Instances. 2023 Asia Meeting on Environment and Electrical Engineering (EEE-AM), 1–5. https://doi.org/10.1109/EEE-AM58328.2023.10394856
Patton, A. J. (2004). On the Out-of-Sample Importance of Skewness and Asymmetric Dependence for Asset Allocation. Journal of Financial Econometrics, 2(1), 130–168. https://doi.org/10.1093/jjfinec/nbh006
Ross, G. J., Adams, N. M., Tasoulis, D. K., & Hand, D. J. (2012). Exponentially weighted moving average charts for detecting concept drift. Pattern Recognition Letters, 33(2), 191–198.
Singh, G., Tripathy, B., & Singh, J. (2023). An Evaluation of Extreme Learning Machine Algorithm to Forecasting the Gold Price. 2023 7th International Conference on Electronics, Communication and Aerospace Technology (ICECA), 523–529. https://api.semanticscholar.org/CorpusID:267576584
Wang, P., Jin, N., Woo, W. L., Woodward, J. R., & Davies, D. (2022). Noise tolerant drift detection method for data stream mining. Information Sciences, 609, 1318–1333. https://doi.org/https://doi.org/10.1016/j.ins.2022.07.065
Wozniak, M., Ksieniewicz, P., Cyganek, B., & Walkowiak, K. (2016). Ensembles of Heterogeneous Concept Drift Detectors - Experimental Study. International Conference on Computer Information Systems and Industrial Management Applications. https://api.semanticscholar.org/CorpusID:2438392
Zhao, P., Cai, L.-W., & Zhou, Z.-H. (2020). Handling concept drift via model reuse. Machine Learning, 109, 533–568. |
| Description: | 碩士
國立政治大學
資訊管理學系
111356008 |
| Source URI: | http://thesis.lib.nccu.edu.tw/record/#G0111356008 |
| Data Type: | thesis |
| Appears in Collections: | [資訊管理學系] 學位論文
|
Files in This Item:
| File |
Description |
Size | Format | |
|---|
| 600801.pdf | | 2561Kb | Adobe PDF | 0 | View/Open |
|
All items in 政大典藏 are protected by copyright, with all rights reserved.
|
