1. NCCU Academic Hub
  2. 學術產出
  3. College of Commerce
  4. Theses
  5. Item

學術產出-Theses

Article View/Open

Publication Export

Google ScholarTM

政大圖書館

Citation Infomation

  • No doi shows Citation Infomation
題名 機器學習資產配置與台股ESG多因子投資組合建構
Machine Learning Asset Allocation and Construction of ESG Multi-Factor Portfolios in the Taiwan Stock Market
作者 林浩詳
Lin, Hau-Siang
貢獻者 羅秉政
Lo, PING-CHENG
林浩詳
Lin, Hau-Siang
關鍵詞 機器學習
超參數
資產配置最適化
多因子投資
股利率因子
獲利因子
動能因子
絕對報酬
ETF
ESG
台股
Machine learning
Hyperparameters
Asset allocation optimization
Multi-factor investment
Dividend yield factor
Profit factor
Momentum factor
Absolute return
ETF
ESG
Taiwanese stocks
日期 2024
上傳時間 1-Mar-2024 12:34:34 (UTC+8)
摘要 本研究主要探討建構台股 ESG 計量投資組合，並應用於投信絕對報酬帳戶或是相對報酬帳戶之操作策略，主要分為以下流程。第一、資產配置:應用機器學習模型預測次月股債指數報酬率與波動率，再將其進行股債效率前緣最適化投資比重配置，輸入全球總經變數、利率、匯率、股價與債券指數、原物料報價等因子 (x_t)，進行機器學習模型訓練，其中包括長短期記憶 （Long Short-Term Memory, LSTM）、循環門單元（Gate Recurrent Unit, GRU）模型。在訓練集 (2005/01至2014/12)優化超參數使股債預測漲跌幅之損失函數最小化，並檢視模型預測值與進行資產配置後績效之穩定度。並於測試集 (2015/01至2019/12)與驗證集 (2020/01至2021/12)觀察與調整。實際應用於真實帳戶 (2022/01至2023/12)。 第二，近年 ESG 議題持續受各界重視，不論是政府勞動基金針對出具企業社會責任(CSR)報告書作為可投資清單外，投信業者亦持續推出 ESG概念 ETF產品，或針對 ESG分數較高的公司，納入股票池，對於有投入 ESG公司股價已產生一定影響力，本研究發現E因子分數逐年提升之公司，將具備較高的夏普值 (Sharpe Ratio)，且在空頭市場表現也相對抗跌。本研究將比照政府勞動基金可投資清單，從中採用近年盛行於台灣指數公司所發行的多因子指數、 Smart Beta策略等方式，建構穩健投資組合，本研究觀察近年台灣股利率因子表現十分優異，若結合獲利因子與動能因子將可再提高超額報酬。 從股票配置比重與挑選股票組成採用計量化投資模式，紀律性建構穩健投資組合。
This paper primarily investigates the construction of an ESG investment portfolio for the Taiwanese stocks. The portfolio is then applied to the operational strategies of mutual funds in absolute or relative return accounts. Asset Allocation: Utilizing machine learning models to predict the next month's stock and bond index returns and volatility. Implementing stock and bond efficiency frontier optimal investment weight allocation based on the predictions. Inputting factors(x_t)such as global macroeconomic variables, interest rates, exchange rates, stock and bond indices, commodity prices, etc., into machine learning models, including LSTM and GRU. Training the models on the training set to optimize hyperparameters and minimize the loss function for predicting stock and bond price movements. Assessing the stability of model predictions and the performance after asset allocation. ESG Impact: Highlighting the impact of ESG considerations on stock prices, as observed in the annual improvement of Environmental (E) factor scores correlating with higher Sharpe Ratios. Recognizing the resilience of companies with increasing ESG scores, even in bear markets. Adapting strategies employed by government labor funds, including the incorporation of recent prevalent multi-factor indices and Smart Beta strategies for constructing robust portfolios. Noting the outstanding performance of the dividend yield factor in the Taiwanese market and the potential for enhanced returns by combining it with profit and momentum factors. The study employs a disciplined quantitative investment approach for asset allocation and stock selection to construct a robust investment portfolio.
參考文獻 [1] 蔡伶婕(2020)。長短期記憶神經網路(LSTM)利率之預測。國立政治大學金融研究所碩士論文，台北市。 [2] Akhigbe, A., & Madura, J. (1996). Dividend policy and corporate performance. Journal of Business Finance & Accounting, 23(9‐10), 1267-1287. [3] Asness, C. S., Moskowitz, T. J., & Pedersen, L. H. (2013). Value and Momentum Everywhere. Journal of Finance, 68(3), 929-985. [4] Amihud, Y., Hameed, A., Kang, W., & Zhang, H. (2015). The illiquidity premium: International evidence. Journal of Financial Economics, 117(2), 350-368. [5] Almahdi, S., & Yang, S. Y. (2017). An adaptive portfolio trading system: A risk-return portfolio optimization using recurrent reinforcement learning with expected maximum drawdown. Expert Systems with Applications, 87, 267-279. [6] Althelaya, K. A., El-Alfy, E.-S. M., & Mohammed, S. (2018). Stock Market Forecast Using Multivariate Analysis with Bidirectional and Stacked (LSTM, GRU). In 2018 21st Saudi Computer Society National Computer Conference (NCC) (pp. 1-7). Riyadh, Saudi Arabia. doi: 10.1109/NCG.2018.8593076. [7] Baskin, J. (1989). Dividend policy and the volatility of common stocks. Journal of portfolio Management, 15(3), 19. [8] Barroso, P., & Santa-Clara, P. (2015). Momentum has its moments. Journal of Financial Economics, 116(1), 111-120. [9] Chordia, T., & Shivakumar, L. (2002). Momentum, Business Cycle, and Time- varying Expected Returns. Journal of Finance, 57(2), 985-1019. [10] Chaves, D., Hsu, J., Li, F., & Shakernia, O. (2011). Risk parity portfolio vs. other asset allocation heuristic portfolios. Journal of Investing. [11] Cho, K., van Merrienboer, B., Gulcehre, C., Bahdanau, D., Bougares, F., Schwenk, H., & Bengio, Y. (2014). Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation. arXiv preprint arXiv:1406.1078. [12] Cahan, E., & Ji, L. (2016, August). Asian Equity Fundamental Factor Model. PORTFOLIO & RISK ANALYTICS Bloomberg. Retrieved from Bloomberg PORT function paper. [13] Cosereanu, C., & Edler, D. (2021, February 16). Emulate Dalio’s Risk-Parity Strategy With Portfolio Optimizer. Bloomberg. Retrieved from Bloomberg FFM function report. [14] Chen, W., Zhang, H., Mehlawat, M. K., & Jia, L. (2021). Mean–variance portfolio optimization using machine learning-based stock price prediction. Applied Soft Computing, 100, 106943. [15] Dimson, E., Karakaş, O., & Li, X. (2015). Active Ownership. The Review of Financial Studies, 28(12), 3225–3268. [16] Eccles, R. G., Ioannou, I., & Serafeim, G. (2014). The Impact of Corporate Sustainability on Organizational Processes and Performance. Management Science, 60(11). [17] Fama, E.F., & French, K.R. (1988). Dividend yields and expected stock returns. Journal of Financial Economics, 22(1), 3-25. [18] Fama, E. F., & French, K. R. (1992). The Cross-Section of Expected Stock Returns. Journal of Finance, 47(2), 427-465. [19] Fama, E.F., & French, K.R. (1993). Common risk factors in the return on stocks and bonds. Journal of Financial Economics, 33(1), 3-56. [20] Fama, E.F., & French, K.R. (1996). Multifactor Explanation of Asset Pricing Anomalies. Journal of Finance, 51(1), 55-84. [21] Flammer, C. (2012). Corporate Social Responsibility and Shareholder Reaction: The Environmental Awareness of Investors. Academy of Management Journal, 56(3). [22] Fisher, G. S., Shah, R., & Titman, S. (2015). Combining value and momentum. Journal of Investment Management, Forthcoming. [23] Friede, G., Busch, T., & Bassen, A. (2015). ESG and financial performance: aggregated evidence from more than 2000 empirical studies. Journal of sustainable finance & investment, 5(4), 210-233. [24] Gordon, M. J. (1959). Dividends, earnings, and stock prices. The review of economics and statistics, 99-105. [25] Gulen, H., Xing, Y., & Zhang, L. (2011). Value versus Growth: Time-Varying Expected Stock Returns. Journal of Empirical Finance, 40(2), 381-407. [26] Hong, H., & Kacperczyk, M. (2009). The price of sin: The effects of social norms on markets. Journal of Financial Economics, 93(1), 15-36. [27] Hou, K., Karolyi, G. A., & Kho, B.‐C. (2011). What Factors Drive Global Stock Returns? The Review of Financial Studies, 24(8), 2527–2574. [28] Hou, K., Xue, C., & Zhang, L. (2015). Digesting anomalies: An investment approach. The Review of Financial Studies, 28(3), 650-705. [29] Heaton, J.B., Polson, N.G., & Witte, J.H. (2016). Deep learning in Finance. arXiv preprint arXiv:1602.06561. [30] Jegadeesh, N., & Titman, S. (1993). Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency.Journal of Finance, 48(1), 65-91. [31] Jegadeesh, N., & Titman, S. (2002). Profitability of Momentum Strategies: An Evaluation of Alternative Explanations. Journal of Finance, 56(2), 699-720. [32] Lakonishok, J., Shleifer, A., & Vishny, R. W. (1994). Contrarian Investment, Extrapolation, and Risk. Journal of Finance, 49(5), 1541-1578. [33] Lee, B. S. (1996). Comovements of earnings, dividends, and stock prices. Journal of Empirical Finance, 3(4), 327-346. [34] Markowitz, H. (1952). Portfolio selection. Journal of Finance, 7(1), 77-91. [35] Moskowitz, T. J., Ooi, Y. H., & Pedersen, L. H. (2012). Time series momentum⋆. Journal of Financial Economics, 104(2), 228-250. [36] Naughton, T., Truong, C., & Veeraraghavan, M. (2008). Momentum strategies and stock returns: Chinese evidence. Pacific-Basin Finance Journal, 16(4), 476-492. [37] Novy-Marx, R. (2012). Is momentum really momentum?. Journal of Financial Economics, 103(3), 429-453. [38] Obeidat, S., Shapiro, D., Lemay, M., MacPherson, M. K., & Bolic, M. (2018). Adaptive Portfolio Asset Allocation Optimization with Deep Learning. International Journal on Advances in Intelligent Systems, 11(1 & 2), 25-34. [39] Paster ,L., & Stambaugh, R. F. (2001). Liquidity Risk and Expected Stock Returns.Journal of Political Economy, 111(3), 642-685. [40] Rouwenhorst, K. G. (1998). International Momentum Strategies. Journal of Finance, 53(1), 267-284. [41] S. Hochreiter and J. Schmidhuber, "Long Short-Term Memory," in Neural Computation, vol. 9, no. 8, pp. 1735-1780, 15 Nov. 1997, doi: 10.1162/neco.1997.9.8.1735.
描述 碩士
國立政治大學
金融學系
108352003
資料來源 http://thesis.lib.nccu.edu.tw/record/#G0108352003
資料類型 thesis
dc.contributor.advisor 羅秉政zh_TW
dc.contributor.advisor Lo, PING-CHENGen_US
dc.contributor.author (Authors) 林浩詳zh_TW
dc.contributor.author (Authors) Lin, Hau-Siangen_US
dc.creator (作者) 林浩詳zh_TW
dc.creator (作者) Lin, Hau-Siangen_US
dc.date (日期) 2024en_US
dc.date.accessioned 1-Mar-2024 12:34:34 (UTC+8)-
dc.date.available 1-Mar-2024 12:34:34 (UTC+8)-
dc.date.issued (上傳時間) 1-Mar-2024 12:34:34 (UTC+8)-
dc.identifier (Other Identifiers) G0108352003en_US
dc.identifier.uri (URI) https://nccur.lib.nccu.edu.tw/handle/140.119/150121-
dc.description (描述) 碩士zh_TW
dc.description (描述) 國立政治大學zh_TW
dc.description (描述) 金融學系zh_TW
dc.description (描述) 108352003zh_TW
dc.description.abstract (摘要) 本研究主要探討建構台股 ESG 計量投資組合，並應用於投信絕對報酬帳戶或是相對報酬帳戶之操作策略，主要分為以下流程。第一、資產配置:應用機器學習模型預測次月股債指數報酬率與波動率，再將其進行股債效率前緣最適化投資比重配置，輸入全球總經變數、利率、匯率、股價與債券指數、原物料報價等因子 (x_t)，進行機器學習模型訓練，其中包括長短期記憶 （Long Short-Term Memory, LSTM）、循環門單元（Gate Recurrent Unit, GRU）模型。在訓練集 (2005/01至2014/12)優化超參數使股債預測漲跌幅之損失函數最小化，並檢視模型預測值與進行資產配置後績效之穩定度。並於測試集 (2015/01至2019/12)與驗證集 (2020/01至2021/12)觀察與調整。實際應用於真實帳戶 (2022/01至2023/12)。 第二，近年 ESG 議題持續受各界重視，不論是政府勞動基金針對出具企業社會責任(CSR)報告書作為可投資清單外，投信業者亦持續推出 ESG概念 ETF產品，或針對 ESG分數較高的公司，納入股票池，對於有投入 ESG公司股價已產生一定影響力，本研究發現E因子分數逐年提升之公司，將具備較高的夏普值 (Sharpe Ratio)，且在空頭市場表現也相對抗跌。本研究將比照政府勞動基金可投資清單，從中採用近年盛行於台灣指數公司所發行的多因子指數、 Smart Beta策略等方式，建構穩健投資組合，本研究觀察近年台灣股利率因子表現十分優異，若結合獲利因子與動能因子將可再提高超額報酬。 從股票配置比重與挑選股票組成採用計量化投資模式，紀律性建構穩健投資組合。zh_TW
dc.description.abstract (摘要) This paper primarily investigates the construction of an ESG investment portfolio for the Taiwanese stocks. The portfolio is then applied to the operational strategies of mutual funds in absolute or relative return accounts. Asset Allocation: Utilizing machine learning models to predict the next month's stock and bond index returns and volatility. Implementing stock and bond efficiency frontier optimal investment weight allocation based on the predictions. Inputting factors(x_t)such as global macroeconomic variables, interest rates, exchange rates, stock and bond indices, commodity prices, etc., into machine learning models, including LSTM and GRU. Training the models on the training set to optimize hyperparameters and minimize the loss function for predicting stock and bond price movements. Assessing the stability of model predictions and the performance after asset allocation. ESG Impact: Highlighting the impact of ESG considerations on stock prices, as observed in the annual improvement of Environmental (E) factor scores correlating with higher Sharpe Ratios. Recognizing the resilience of companies with increasing ESG scores, even in bear markets. Adapting strategies employed by government labor funds, including the incorporation of recent prevalent multi-factor indices and Smart Beta strategies for constructing robust portfolios. Noting the outstanding performance of the dividend yield factor in the Taiwanese market and the potential for enhanced returns by combining it with profit and momentum factors. The study employs a disciplined quantitative investment approach for asset allocation and stock selection to construct a robust investment portfolio.en_US
dc.description.tableofcontents 第一章 緒論 1 第一節 研究背景及動機 1 第二節 研究目的 2 第三節 論文架構及章節介紹 3 第二章 文獻回顧 4 第一節 股利因子文獻探討 4 第二節 動能因子文獻探討 5 第三節 價值因子文獻探討 7 第四節 ESG因子文獻探討 9 第五節 多因子文獻探討 10 第六節 機器學習與資產配置文獻探討 12 第三章 資產配置研究方法 15 第一節 研究架構 15 第二節 採用機器學習進行資產配置 17 第三節 常見配置模型與其績效比較 25 第四章 多因子投資研究方法 31 第一節 研究架構 31 第二節 CSR股票池與流動性考量 32 第三節 多因子類型、特性與選股策略 33 第四節 環境因子與社會因子優化多因子權重 40 第五章 實證結果分析結論與未來建議 42 參考文獻 45 附錄一 機器學習資產配置模型輸入變數 49 附錄二 多因子策略因子池 60 附錄三 多因子策略最新投資組合 68zh_TW
dc.source.uri (資料來源) http://thesis.lib.nccu.edu.tw/record/#G0108352003en_US
dc.subject (關鍵詞) 機器學習zh_TW
dc.subject (關鍵詞) 超參數zh_TW
dc.subject (關鍵詞) 資產配置最適化zh_TW
dc.subject (關鍵詞) 多因子投資zh_TW
dc.subject (關鍵詞) 股利率因子zh_TW
dc.subject (關鍵詞) 獲利因子zh_TW
dc.subject (關鍵詞) 動能因子zh_TW
dc.subject (關鍵詞) 絕對報酬zh_TW
dc.subject (關鍵詞) ETFzh_TW
dc.subject (關鍵詞) ESGzh_TW
dc.subject (關鍵詞) 台股zh_TW
dc.subject (關鍵詞) Machine learningen_US
dc.subject (關鍵詞) Hyperparametersen_US
dc.subject (關鍵詞) Asset allocation optimizationen_US
dc.subject (關鍵詞) Multi-factor investmenten_US
dc.subject (關鍵詞) Dividend yield factoren_US
dc.subject (關鍵詞) Profit factoren_US
dc.subject (關鍵詞) Momentum factoren_US
dc.subject (關鍵詞) Absolute returnen_US
dc.subject (關鍵詞) ETFen_US
dc.subject (關鍵詞) ESGen_US
dc.subject (關鍵詞) Taiwanese stocksen_US
dc.title (題名) 機器學習資產配置與台股ESG多因子投資組合建構zh_TW
dc.title (題名) Machine Learning Asset Allocation and Construction of ESG Multi-Factor Portfolios in the Taiwan Stock Marketen_US
dc.type (資料類型) thesisen_US
dc.relation.reference (參考文獻) [1] 蔡伶婕(2020)。長短期記憶神經網路(LSTM)利率之預測。國立政治大學金融研究所碩士論文，台北市。 [2] Akhigbe, A., & Madura, J. (1996). Dividend policy and corporate performance. Journal of Business Finance & Accounting, 23(9‐10), 1267-1287. [3] Asness, C. S., Moskowitz, T. J., & Pedersen, L. H. (2013). Value and Momentum Everywhere. Journal of Finance, 68(3), 929-985. [4] Amihud, Y., Hameed, A., Kang, W., & Zhang, H. (2015). The illiquidity premium: International evidence. Journal of Financial Economics, 117(2), 350-368. [5] Almahdi, S., & Yang, S. Y. (2017). An adaptive portfolio trading system: A risk-return portfolio optimization using recurrent reinforcement learning with expected maximum drawdown. Expert Systems with Applications, 87, 267-279. [6] Althelaya, K. A., El-Alfy, E.-S. M., & Mohammed, S. (2018). Stock Market Forecast Using Multivariate Analysis with Bidirectional and Stacked (LSTM, GRU). In 2018 21st Saudi Computer Society National Computer Conference (NCC) (pp. 1-7). Riyadh, Saudi Arabia. doi: 10.1109/NCG.2018.8593076. [7] Baskin, J. (1989). Dividend policy and the volatility of common stocks. Journal of portfolio Management, 15(3), 19. [8] Barroso, P., & Santa-Clara, P. (2015). Momentum has its moments. Journal of Financial Economics, 116(1), 111-120. [9] Chordia, T., & Shivakumar, L. (2002). Momentum, Business Cycle, and Time- varying Expected Returns. Journal of Finance, 57(2), 985-1019. [10] Chaves, D., Hsu, J., Li, F., & Shakernia, O. (2011). Risk parity portfolio vs. other asset allocation heuristic portfolios. Journal of Investing. [11] Cho, K., van Merrienboer, B., Gulcehre, C., Bahdanau, D., Bougares, F., Schwenk, H., & Bengio, Y. (2014). Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation. arXiv preprint arXiv:1406.1078. [12] Cahan, E., & Ji, L. (2016, August). Asian Equity Fundamental Factor Model. PORTFOLIO & RISK ANALYTICS Bloomberg. Retrieved from Bloomberg PORT function paper. [13] Cosereanu, C., & Edler, D. (2021, February 16). Emulate Dalio’s Risk-Parity Strategy With Portfolio Optimizer. Bloomberg. Retrieved from Bloomberg FFM function report. [14] Chen, W., Zhang, H., Mehlawat, M. K., & Jia, L. (2021). Mean–variance portfolio optimization using machine learning-based stock price prediction. Applied Soft Computing, 100, 106943. [15] Dimson, E., Karakaş, O., & Li, X. (2015). Active Ownership. The Review of Financial Studies, 28(12), 3225–3268. [16] Eccles, R. G., Ioannou, I., & Serafeim, G. (2014). The Impact of Corporate Sustainability on Organizational Processes and Performance. Management Science, 60(11). [17] Fama, E.F., & French, K.R. (1988). Dividend yields and expected stock returns. Journal of Financial Economics, 22(1), 3-25. [18] Fama, E. F., & French, K. R. (1992). The Cross-Section of Expected Stock Returns. Journal of Finance, 47(2), 427-465. [19] Fama, E.F., & French, K.R. (1993). Common risk factors in the return on stocks and bonds. Journal of Financial Economics, 33(1), 3-56. [20] Fama, E.F., & French, K.R. (1996). Multifactor Explanation of Asset Pricing Anomalies. Journal of Finance, 51(1), 55-84. [21] Flammer, C. (2012). Corporate Social Responsibility and Shareholder Reaction: The Environmental Awareness of Investors. Academy of Management Journal, 56(3). [22] Fisher, G. S., Shah, R., & Titman, S. (2015). Combining value and momentum. Journal of Investment Management, Forthcoming. [23] Friede, G., Busch, T., & Bassen, A. (2015). ESG and financial performance: aggregated evidence from more than 2000 empirical studies. Journal of sustainable finance & investment, 5(4), 210-233. [24] Gordon, M. J. (1959). Dividends, earnings, and stock prices. The review of economics and statistics, 99-105. [25] Gulen, H., Xing, Y., & Zhang, L. (2011). Value versus Growth: Time-Varying Expected Stock Returns. Journal of Empirical Finance, 40(2), 381-407. [26] Hong, H., & Kacperczyk, M. (2009). The price of sin: The effects of social norms on markets. Journal of Financial Economics, 93(1), 15-36. [27] Hou, K., Karolyi, G. A., & Kho, B.‐C. (2011). What Factors Drive Global Stock Returns? The Review of Financial Studies, 24(8), 2527–2574. [28] Hou, K., Xue, C., & Zhang, L. (2015). Digesting anomalies: An investment approach. The Review of Financial Studies, 28(3), 650-705. [29] Heaton, J.B., Polson, N.G., & Witte, J.H. (2016). Deep learning in Finance. arXiv preprint arXiv:1602.06561. [30] Jegadeesh, N., & Titman, S. (1993). Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency.Journal of Finance, 48(1), 65-91. [31] Jegadeesh, N., & Titman, S. (2002). Profitability of Momentum Strategies: An Evaluation of Alternative Explanations. Journal of Finance, 56(2), 699-720. [32] Lakonishok, J., Shleifer, A., & Vishny, R. W. (1994). Contrarian Investment, Extrapolation, and Risk. Journal of Finance, 49(5), 1541-1578. [33] Lee, B. S. (1996). Comovements of earnings, dividends, and stock prices. Journal of Empirical Finance, 3(4), 327-346. [34] Markowitz, H. (1952). Portfolio selection. Journal of Finance, 7(1), 77-91. [35] Moskowitz, T. J., Ooi, Y. H., & Pedersen, L. H. (2012). Time series momentum⋆. Journal of Financial Economics, 104(2), 228-250. [36] Naughton, T., Truong, C., & Veeraraghavan, M. (2008). Momentum strategies and stock returns: Chinese evidence. Pacific-Basin Finance Journal, 16(4), 476-492. [37] Novy-Marx, R. (2012). Is momentum really momentum?. Journal of Financial Economics, 103(3), 429-453. [38] Obeidat, S., Shapiro, D., Lemay, M., MacPherson, M. K., & Bolic, M. (2018). Adaptive Portfolio Asset Allocation Optimization with Deep Learning. International Journal on Advances in Intelligent Systems, 11(1 & 2), 25-34. [39] Paster ,L., & Stambaugh, R. F. (2001). Liquidity Risk and Expected Stock Returns.Journal of Political Economy, 111(3), 642-685. [40] Rouwenhorst, K. G. (1998). International Momentum Strategies. Journal of Finance, 53(1), 267-284. [41] S. Hochreiter and J. Schmidhuber, "Long Short-Term Memory," in Neural Computation, vol. 9, no. 8, pp. 1735-1780, 15 Nov. 1997, doi: 10.1162/neco.1997.9.8.1735.zh_TW