机器学习解构区域金融风险防控研究进展

doi:10.3778/j.issn.1673-9418.2203127

计算机科学与探索 ›› 2022, Vol. 16 ›› Issue (9): 1969-1989.DOI: 10.3778/j.issn.1673-9418.2203127

机器学习解构区域金融风险防控研究进展

张立华¹, 张顺顺²^,⁺()

1.温州商学院金融贸易学院,浙江温州 325000
2.伦敦大学国王学院国王商学院,伦敦 WC2B 4BG

收稿日期:2022-03-30 修回日期:2022-06-08 出版日期:2022-09-01 发布日期:2022-09-15
通讯作者: + E-mail: 1504886520@qq.com
作者简介:张立华（1963—）,男,北京人,博士,教授,主要研究方向为机器学习、金融大数据、金融风险建模。
张顺顺（1996—）,女,北京人,博士研究生,主要研究方向为行为金融、机器学习文本分析、金融大数据。
基金资助:
国家社会科学基金重大项目(18ZDA093)

Progress on Machine Learning for Regional Financial Risk Prevention

ZHANG Lihua¹, ZHANG Shunshun²^,⁺()

1. School of Finance and Trade, Wenzhou Business College, Wenzhou, Zhejiang 325000, China
2. King's Business School, King's College London, London WC2B 4BG, UK

Received:2022-03-30 Revised:2022-06-08 Online:2022-09-01 Published:2022-09-15
About author:ZHANG Lihua, born in 1963, Ph.D., professor. His research interests include machine learning,financial big data and financial risks modelling.
ZHANG Shunshun, born in 1996, Ph.D. candidate. Her research interests include behavioral finance, machine-learning textual analysis and financial big data.
Supported by:
Major Project of the National Social Science Foundation of China(18ZDA093)

摘要/Abstract

摘要：

区域金融风险防控（RFRP）无论在管理区域传统金融风险（TFR）还是坚守不发生区域金融系统风险（FSR）中都是不可或缺的。随着大数据规模的持续增长,金融风险形态变化的不确定性,传统计量方法模拟金融风险防控的效率、精度、应用等方面都面临着无法克服的困境。当下,越来越多的机器学习（ML）模拟RFRP防控的新方法和新技术受到研究者的重视。首先提出了RFRP防控新的科学分类和ML观念基础;其次总结了区域TFR防控的ML理论方法和应用技术,对各类代表性研究所论述区域TFR防控的关键逻辑、模型算法、学习效果进行了比对解析,对ML不同方法的优点、局限和传统场景进行了归类分析;然后梳理了区域FSR防控的ML理论方法和应用研究,对各类典型文献所解析区域FSR防控的关键脉络、ML算法、学习效果进行了对比研究,对ML不同模型的优势、缺陷和金融风险场景进行了阐述研究;最后提出了六个ML模拟RFRP防控的前景技术和新兴方向。

关键词: 机器学习（ML）, 区域金融风险防控（RFRP）, 传统金融风险（TFR）, 金融系统风险（FSR）

Abstract:

The regional financial risks prevention (RFRP) is indispensable in managing the regional traditional financial risks (TFR) or preventing the regional financial systemic risks (FSR). With the growing of big data and the uncertainty of financial risk types, traditional econometrics methods are facing insurmountable difficulties in terms of the efficiency, accuracy, and application of financial risk prevention modeling. Today, increasing future methods and technologies for machine learning (ML) for RFRP prevention have been paid much attention by researchers. A new scientific classification of RFRP prevention and conceptual basis of ML methods are firstly put forward.Secondly, this paper summarizes the ML methods and application of regional TFR prevention, compares their key logic, model algorithm and learning effect of the representative literature, and categorizes the advantages, limitations and traditional scenarios of ML methods. Thirdly, it combs the ML methods and application of regional FSR prevention, analyzes the key context, ML algorithm and learning effect of the seminal documents, and compares the benefits, disadvantages and financial risk contexts of ML methods. Finally, six promising technologies and emerging directions of ML methods for RFRP prevention are proposed.

Key words: machine learning (ML), regional financial risk prevention (RFRP), traditional financial risk (TFR), financial systemic risk (FSR)

中图分类号:

TP18
F83

张立华, 张顺顺. 机器学习解构区域金融风险防控研究进展[J]. 计算机科学与探索, 2022, 16(9): 1969-1989.

ZHANG Lihua, ZHANG Shunshun. Progress on Machine Learning for Regional Financial Risk Prevention[J]. Journal of Frontiers of Computer Science and Technology, 2022, 16(9): 1969-1989.

图/表 21

参考文献 132

[1]	CAVALCANTE R C, BRASILEIRO R C, SOUZA V L, et al. Computational intelligence and financial markets: a survey and future directions[J]. Expert Systems with Applications, 2016, 55: 194-211. DOI URL
[2]	DE SPIEGELEER J, MADAN D B, REYNERS S, et al. Machine learning for quantitative finance: fast derivative pricing, hedging and fitting[J]. Quantitative Finance, 2018, 18(10): 1635-1643. DOI URL
[3]	MASHRUR A, LUO W, ZAIDI N A, et al. Machine learning for financial risk management: a survey[J]. IEEE Access, 2020, 8: 203203-203223. DOI URL
[4]	HELBING D. Globally networked risks and how to respond[J]. Nature, 2013, 497: 52-59.
[5]	KUO K. DeepTriangle: a deep learning approach to loss re-serving[J]. Risks, 2018, 7(3): 1-12. DOI URL
[6]	GIUDICI P, SARLIN P, SPELTA A. The interconnected nature of financial systems: direct and common exposures[J]. Journal of Banking & Finance, 2020, 112: 105149. DOI URL
[7]	SILVA W, KIMURA H, SOBREIRO V A. An analysis of the literature on systemic financial risk: a survey[J]. Journal of Financial Stability, 2017, 28: 91-114. DOI URL
[8]	张立华, 丁建臣. 高阶矩、HCCA模型与银行系统风险前瞻预判[J]. 统计研究, 2016, 33(1): 70-77.
	ZHANG L H, DING J C. Prospective prediction of banking systemic risk with HCCA model[J]. Statistical Research, 2016, 33(1): 70-77.
[9]	张立华, 张顺顺. 系统性风险与系统风险[J]. 中国金融, 2017(8): 95-96.
	ZHANG L H, ZHANG S S. Systematic risk and systemic risk[J]. China Finance, 2017(8): 95-96.
[10]	KUBAT M. An introduction to machine learning[M]. Berlin, Heidelberg: Springer, 2017.
[11]	GOLDSTEIN M, UCHIDA S. A comparative evaluation of unsupervised anomaly detection algorithms for multivariate data[J]. PLoS One, 2016, 11(4): e0152173. DOI URL
[12]	VAN ENGELEN J E, HOOS H H. A survey on semisuper-vised learning[J]. Machine Learning, 2020, 109: 373-440. DOI URL
[13]	MONDAL A K. A survey of reinforcement learning techni-ques: strategies, recent development, and future directions[J]. arXiv:2001.06921v2, 2020.
[14]	GOODFELLOW I, BENGIO Y, COURVILLE A. Deep learning[M]. Cambridge: MIT Press, 2016.
[15]	GUARASCIO M, MANCO G, RITACCO E. Deep learning[M]// RANGANATHANS, GRIBSKOVM,NAKAI K, eds.eds. Encyclopedia of Bioinformatics and Computational Bio-logy. New York: Elsevier Science Inc., 2019: 634-647.
[16]	LECUN Y, BENGIO Y, HINTON G. Deep learning[J]. Nature, 2015, 521(7553): 436-444. DOI URL
[17]	LIM B, ZOHREN S. Time series forecasting with deep learning: a survey[J]. arXiv:2004.13408v2, 2020.
[18]	LIN W Y, HU Y H, TSAI C F. Machine learning in financial crisis prediction: a survey[J]. IEEE Transactions on Sys-tems, Man, and Cybernetics, Part C, 2012, 42(4): 421-436.
[19]	CHEN M Y. Bankruptcy prediction in firms with statistical and intelligent techniques and a comparison of evolution-ary computation approaches[J]. Computers and Mathematics with Applications, 2011, 62 (12): 4514-4524. DOI URL
[20]	WEST D. Neural network credit scoring models[J]. Compu-ters & Operations Research, 2000, 27: 1131-1152.
[21]	KIM K J, AHN H. A corporate credit rating model using multi-class support vector machines with an ordinal pairwise partitioning approach[J]. Computers & Operations Research, 2012, 39(8): 1800-1811. DOI URL
[22]	KIM S Y, UPNEJA A. Predicting restaurant financial distress using decision tree and AdaBoosted decision tree models[J]. Economic Modelling, 2014, 36: 354-362. DOI URL
[23]	BARBOZA F, KIMURA H, ALTMAN E. Machine learning models and bankruptcy prediction[J]. Expert Systems with Applications, 2017, 83: 405-417. DOI URL
[24]	GHODSELAHI A. A hybrid support vector machine ense-mble model for credit scoring[J]. International Journal of Computer Applications, 2011, 17(5): 1-5.
[25]	HU L, CHEN J, VAUGHAN J, et al. Supervised machine learning techniques: an overview with applications to ban-king[J]. arXiv:2008.04059v1, 2020.
[26]	WANG G, MA J. A hybrid ensemble approach for enterprise credit risk assessment based on support vector machine[J]. Expert Systems with Applications, 2012, 39(5): 5325-5331. DOI URL
[27]	AMPOUNTOLAS A, NDE T N, DATE P, et al. A machine learning approach for micro-credit scoring[J]. Risks, 2021, 9(3): 50. DOI URL
[28]	HARDING M, VASCONCELOS G. Managers versus mac-hines: do algorithms replicate human intuition in credit ratings?[J]. arXiv:2202.04218v1, 2022.
[29]	LIM M K, SOHNS Y. Cluster-based dynamic scoring model[J]. Expert Systems with Applications, 2007, 32(2): 427-431. DOI URL
[30]	KOU G, PENG Y, WANG G. Evaluation of clustering algori-thms for financial risk analysis using MCDM methods[J]. Information Sciences, 2014, 275: 1-12. DOI URL
[31]	SHILLER R J. Market volatility[M]. Cambridge: MIT Press, 1992.
[32]	LUONG C, DOKUCHAEV N. Forecasting of realised vola-tility with the random forests algorithm[J]. Journal of Risk and Financial Management, 2018, 11(4): 1-15. DOI URL
[33]	PARK H, KIM N, LEE J. Parametric models and non-parametric machine learning models for predicting option prices: empirical comparison study over KOSPI 200 index options[J]. Expert Systems with Applications, 2014, 41(11): 5227-5237. DOI URL
[34]	LUO R, ZHANG W N, XU X J, et al. A neural stochastic volatility model[C]// Proceedings of the 32nd AAAI Con-ference on Artificial Intelligence, the 30th Innovative App-lications of Artificial Intelligence, and the 8th AAAI Sym-posium on Educational Advances in Artificial Intelligence, New Orleans, Feb 2-7, 2018. Menlo Park: AAAI, 2018: 6401-6408.
[35]	HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780. DOI URL
[36]	RAMOS-PÉREZ E, ALONSO-GONZÁLEZ P J, NÚEZ-VELÁZQUEZ J J, et al. Forecasting volatility with a stacked model based on a hybridized artificial neural network[J]. arXiv:2006.16383v2, 2020.
[37]	GROTH S S, MUNTERMANN J. An intraday market risk management approach based on textual analysis[J]. Decision Support Systems, 2011, 50(4): 680-691. DOI URL
[38]	ANTWEILER W, FRANK M Z. Is all that talk just noise? The information content of Internet stock message boards[J]. The Journal of Finance, 2004, 59(3): 1259-1294. DOI URL
[39]	NIZER P S M, NIEVOLAJ C. Predicting published news effect in the Brazilian stock market[J]. Expert Systems with Applications, 2012, 39(12): 10674-10680. DOI URL
[40]	MANELA A, MOREIRA A. News implied volatility and disaster concerns[J]. Journal of Financial Economics, 2017, 23(1): 137-162.
[41]	OLIVEIRA N, CORTEZ P, AREAL N. The impact of micro-blogging data for stock market prediction: using Twitter to predict returns, volatility, trading volume and survey senti-ment indices[J]. Expert Systems with Applications, 2017, 73: 125-144. DOI URL
[42]	XU Q, BO Z, JIANG C, et al. Does Google search index really help predicting stock market volatility? Evidence from a modified mixed data sampling model on volatility[J]. Knowledge-Based Systems, 2019, 166: 170-185. DOI URL
[43]	XU Y M, COHEN S B. Stock movement prediction from tweets and historical prices[C]// Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics, Melbourne, Jul 15-20, 2018. Stroudsburg: ACL, 2018: 1970-1979.
[44]	PAPADIMITRIOU T, GOGAS P, ATHANASIOU A F. Fore-casting S&P 500 spikes: an SVM approach[J]. Digital Finance, 2020. DOI: 10.1007/s42521-020-00024-0. DOI
[45]	BAN G Y, KAROUIN E, LIMA E. Machine learning and portfolio optimization[J]. Management Science, 2018, 64(3): 1136-1154. DOI URL
[46]	PINELIS M, RUPPERT D. Machine learning portfolio allo-cation[J]. arXiv:2003.00656v2, 2020.
[47]	LI B, ZHAO P, HOI S C H, et al. PAMR: passive aggressive mean reversion strategy for portfolio selection[J]. Machine Learning, 2012, 87(2): 221-258. DOI URL
[48]	LI B, HOI S C. Online portfolio selection: a survey[J]. ACM Computing Surveys, 2014, 46(3): 1-36.
[49]	JIANG Z, XU D, LIANG J. A deep reinforcement learning framework for the financial portfolio management problem[J]. arXiv:1706.10059, 2017.
[50]	ALMAHDI S, YANGS Y. An adaptive portfolio trading sys-tem: a risk return portfolio optimization using recurrent reinforcement learning with expected maximum drawdown[J]. Expert Systems with Applications, 2017, 87: 267-279. DOI URL
[51]	NGAI E, HU Y, WONG Y H, et al. The application of data mining techniques in financial fraud detection: a classifi-cation framework and an academic review of literature[J]. Decision Support Systems, 2011, 50(3): 559-569. DOI URL
[52]	SAHIN Y, BULKAN S, DUMAN E. A cost-sensitive deci-sion tree approach for fraud detection[J]. Expert Systems with Applications, 2013, 40(15): 5916-5923. DOI URL
[53]	BAHNSEN A C, AOUADA D, STOJANOVIC A, et al. Fea-ture engineering strategies for credit card fraud detection[J]. Expert Systems with Applications, 2016, 51: 134-142. DOI URL
[54]	ABBASI A, ALBRECHT C, VANCE A, et al. MetaFraud: a metalearning framework for detecting financial fraud[J]. MIS Quarterly, 2012, 36(4): 1293-1327. DOI URL
[55]	KUSAYA C, ANALYST R, O'KEEFE J. Insider abuse and fraud prediction for U.S. banks: a comparison of machine lear-ning approaches[J]. SSRN Electronic Journal, 2021. DOI: 10.2139/ssrn.3659757. DOI
[56]	LOUGHRAN T, MCDONALD B. Textual analysis in accou-nting and finance: a survey[J]. Journal of Accounting Re-search, 2016, 54(4): 1187-1230.
[57]	NIAN K, ZHANG H, TAYAL A, et al. Auto insurance fraud detection using unsupervised spectral ranking for anomaly[J]. The Journal of Finance and Data Science, 2016, 2(1): 58-75. DOI URL
[58]	PAPERNOT N, MCDANIEL P, GOODFELLOW I. Trans-ferability in machine learning: from phenomena to black-box attacks using adversarial samples[J]. arXiv:1605.07277, 2016.
[59]	CHALAPATHY R, CHAWLA S. Deep learning for anomaly detection: a survey[J]. arXiv:1901.03407, 2019.
[60]	FIORE U, DE SANTIS A, PERLA F, et al. Using generative adversarial networks for improving classification effective-ness in credit card fraud detection[J]. Information Sciences, 2019, 479: 448-455. DOI URL
[61]	PURDA L, SKILLICORN D. Accounting variables, decep-tion, and a bag of words: assessing the tools of fraud detec-tion[J]. Contemporary Accounting Research, 2015, 32(3): 1193-1223. DOI URL
[62]	HAJEK P, HENRIQUES R. Mining corporate annual reports for intelligent detection of financial statement fraud: a com-parative study of machine learning methods[J]. Knowledge-Based Systems, 2017, 128: 139-152. DOI URL
[63]	VERBELEN R, ANTONIO K, CLAESKENS G. Unravelling the predictive power of telematics data in car insurance pricing[J]. Journal of the Royal Statistical Society, 2018, 67(5): 1275-1304.
[64]	TSAI R Y, HUANG T S. Multiframe image restoration and registration[J]. Advance Computer Visual and Image Pro-cessing, 1984, 1: 317-339.
[65]	GHAFFAR M, MCKINSTRY A, MAUL T, et al. Data aug-mentation approaches for satellite image super-resolution[J]. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 2019: 47-54.
[66]	LEDIG C, THEIS L, HUSZAR F, et al. Photo-realistic single image super-resolution using a generative adversarial net-work[C]// Proceedings of the 2017 IEEE Conference on Com-puter Vision and Pattern Recognition, Honolulu, Jul 21-26, 2017. Washington: IEEE Computer Society, 2017: 105-114.
[67]	LI H, ZHENG Q, YAN W, et al. Image super-resolution re-construction for secure data transmission in Internet of things environment[J]. Mathematical Biosciences and Engineering, 2021, 18(5): 6652-6671. DOI URL
[68]	LEE R D, CARTER L R. Modeling and forecasting US mor-tality[J]. Journal of the American Statistical Association, 1992, 87(419): 659-671.
[69]	MILEVSKY M A, PROMISLOW S D. Mortality derivatives and the option to annuitise[J]. Insurance: Mathematics and Economics, 2001, 29(3): 299-318. DOI URL
[70]	RICHMAN R, WÜTHRICH M V. A neural network exten-sion of the Lee-Carter model to multiple populations[J]. Annals of Actuarial Science, 2021, 15(2): 346-366. DOI URL
[71]	HAINAUT D. A neural-network analyzer for mortality fore-cast[J]. Astin Bulletin, 2018, 48(2): 481-508. DOI URL
[72]	KARTHIKEYAN A, GARG A, VINOD P K, et al. Machine-learning based clinical decision support system for early COVID-19 mortality prediction[J]. Frontiers in Public Health, 2021, 9: 626697. DOI URL
[73]	SRIRAM A, MUCKLEY M, SINHA K, et al. COVID-19 deterioration prediction via self-supervised representation learning and multi-image prediction[J]. arXiv:2101.04909, 2021.
[74]	GOURDEAU D, POTVIN O, ARCHAMBAULT P, et al. Tracking and predicting COVID-19 radiological trajectory on chest X-rays using deep learning[J]. Scientific Reports, 2022, 12(1): 5616. DOI URL
[75]	BATTISTON S, FARMER J D, FLACHE A. Complexity theory and financial regulation[J]. Science, 2016, 351(6275): 818-819. DOI URL
[76]	HALDANE A, MAY R. Systemic risk in banking ecosystems[J]. Nature, 2011, 469(7330): 351-355. DOI URL
[77]	PRASANNA G, HALDANE A, KAPADIA S. Complexity, concentration and contagion[J]. Journal of Monetary Eco-nomics, 2011, 58(5): 453-470.
[78]	HU D, ZHAO J L, HUA Z, et al. Network based modeling and analysis of systemic risk in banking systems[J]. MIS Quarterly, 2012, 36(4): 1269-1291. DOI URL
[79]	DIEBOLD F X, YILMAZ K. On the network topology of variance decompositions: measuring the connectedness of financial firms[J]. Journal of Econometrics, 2014, 182(1): 119-134. DOI URL
[80]	ACEMOGLU D, OZDAGLAR A, TAHBAZ-SALEHI A. Systemic risk and stability in financial networks[J]. Ame-rican Economic Review, 2015, 105(2): 564-608.
[81]	AMINI H, CONT R, MINCA A. Resilience to contagion in financial networks[J]. Mathematical Finance, 2016, 26(2): 329-365. DOI URL
[82]	BLUHM M, KRAHNEN J. Systemic risk in an interconne-cted banking system with endogenous asset markets[J]. Journal of Financial Stability, 2014, 13(1): 75-94. DOI URL
[83]	CHOI D. Heterogeneity and stability: bolster the strong, not the weak[J]. The Review of Financial Studies, 2014, 27(6): 1830-1867. DOI URL
[84]	GIUDICI P, SPELTA A. Graphical network models for inter-national financial flows[J]. Journal of Business & Economic Statistics, 2016, 34(1): 128-138. DOI URL
[85]	YU J, ZHAO J. Prediction of systemic risk contagion based on a dynamic complex network model using machine learning algorithm[J]. Complexity, 2020, 2: 1-13.
[86]	BETZ F, HAUTSCH N, PELTONEN T A, et al. Systemic risk spillovers in the European banking and sovereign net-work[J]. Journal of Financial Stability, 2016, 25: 206-224. DOI URL
[87]	SHEN C W. A Bayesian networks approach to modeling financial risks of e-logistics investments[J]. International Journal of Information Technology & Decision Making, 2017, 8(4): 711-726.
[88]	BATTISTON S, CALDARELLI G, MAY R M, et al. The price of complexity in financial networks[J]. Proceedings of the National Academy of Sciencesof the United States of America, 2016, 113(36): 10-31.
[89]	POLEDNA S, MOLINA-BORBOA J L, MARTÍNEZ-JARAMILLO S, et al. The multi-layer network nature of systemic risk and its implications for the costs of financial crises[J]. Journal of Financial Stability, 2015, 20: 70-81. DOI URL
[90]	GRASSIA M, DE DOMENICO M, MANGIONI G. Machine learning dismantling and early-warning signals of disinte-gration in complex systems[J]. Nature Communications, 2021, 12(1): 5190. DOI URL
[91]	CERCHIELLO P, GIUDICI P. Big data analysis for financial risk management[J]. Journal of Big Data, 2016, 3(1): 18. DOI URL
[92]	SARLIN P. Computational tools for systemic risk identifica-tion and assessment[J]. Intelligent Systems in Accounting, Finance and Management, 2016, 23(1/2): 3-5. DOI URL
[93]	SARLIN P. Macroprudential oversight, risk communication and visualization[J]. Journal of Financial Stability, 2016, 27: 160-179. DOI URL
[94]	CERCHIELLO P, GIUDICI P, NICOLA D. Big data models of bank risk contagion[J]. DEM Working Paper Series, 2016: 117.
[95]	FLOOD M, JAGADISH H V, KYLE A, et al. Using data for systemic financial risk management[C]// Proceedings of the 5th Biennial Conference on Innovative Data Systems Re-search, Asilomar, Jan 9-12, 2011: 144-147.
[96]	JAGADISH H V, GEHRKE J, LABRINIDIS A, et al. Big data and its technical challenges[J]. Communications of the ACM, 2014, 57(7): 86-94.
[97]	YANGS, GUO K, LI J P, et al. Framework formation of financial data classification standard in the era of the big data[J]. Procedia Computer Science, 2014, 30: 88-96. DOI URL
[98]	BRAMMERTZ W, MENDELOWITZA I. Limits and oppor-tunities of big data for macro-prudential modeling of fina-ncial systemic risk[C]// Proceedings of the 2014 Interna-tional Workshop on Data Science for Macro-Modeling, Sno-wbird, Jun 22-27, 2014. New York: ACM, 2014: 1-6.
[99]	NYMAN R, KAPADIA S, TUCKETT D. News and narra-tives in financial systems: exploiting big data for systemic risk assessment[J]. Journal of Economic Dynamics and Control, 2021, 4: 104119.
[100]	CHIANG J K, CHEN C C. Sentimental analysis on big data- on case of financial document text mining to predict sub-index trend[C]// Proceedings of the 5th International Con-ference on Computer Sciences and Automation Engineering, Sanya, Nov 14-15, 2015. Paris: Atlantis Press, 2016: 423-428.
[101]	TSAI M F, WANG C J. On the risk prediction and analysis of soft information in finance reports[J]. European Journal of Operational Research, 2017, 257(1): 243-250. DOI URL
[102]	MEYER B, BIKDASH M, DAI X. Fine-grained financial news sentiment analysis[C]// Proceedings of the Southeast Conference 2017, Charlotte, Mar 30-Apr 2, 2017. Piscataway: IEEE, 2017: 1-8.
[103]	TSAI M F, WANG C J. Risk ranking from financial reports[C]// LNCS 7814: Proceedings of the 35th European Con-ference on IR Research, Moscow, Mar 24-27, 2013. Berlin, Heidelberg: Springer, 2013: 804-807.
[104]	DUTTAGUPTA R, CASHIN P. Anatomy of banking crises in developing and emerging market countries[J]. Journal of International Money and Finance, 2011, 30(2): 354-376. DOI URL
[105]	MANASSE P, SAVONA R, VEZZOLI M. Danger zones for banking crises in emerging markets[J]. International Journal of Finance & Economics, 2016, 21: 360-381.
[106]	GABRIELE C. Learning from trees: a mixed approach to building early warning systems for systemic banking crises[J]. European Stability Mechanism Working Paper, 2019: 40.
[107]	TANAKA K, KINKYO T, HAMORI S. Random forests-based early warning system for bank failures[J]. Economics Letters, 2016, 148: 118-121. DOI URL
[108]	WARD F. Spotting the danger zone: forecasting financial crises with classification tree ensembles and many predi-ctors[J]. Journal of Applied Econometrics, 2017, 32(2): 359-378. DOI URL
[109]	CASABIANCA E J, CATALANO M, FORNI L, et al. An early warning system for banking crises: from regression-based analysis to machine learning techniques[J]. Marco Fanno Working Papers, 2019: 235.
[110]	BLUWSTEIN K, BUCKMANN M, JOSEPH A, et al. Credit growth, the yield curve and financial crisis prediction: evi-dence from a machine learning approach[J]. ECB Working Paper, 2021: 2614.
[111]	ARAKELIAN V, DELLAPORTAS P, SAVONA R, et al. Sovereign risk zones in Europe during and after the debt crisis[J]. Quantitative Finance, 2019, 19(6): 961-980. DOI URL
[112]	JOY M, RUSNAK M, SMIDKOVA K, et al. Banking and currency crises: differential diagnostics for developed coun-tries[J]. International Journal of Finance & Economics, 2017, 22(1): 44-67.
[113]	BARRY C B. Portfolio analysis under uncertain means, va-riances, and covariances[J]. Journal of Finance, 1974, 29: 515-522. DOI URL
[114]	BRODIE J, DAUBECHIES I, MOL C D, et al. Sparse and stable Markowitz portfolios[J]. Proceedings of the National Academy of Sciences of the United States of America, 2009, 106(30): 12267-12272.
[115]	FAN J, ZHANG J, YU K. Vast portfolio selection with gross-exposure constraints[J]. Journal of the American Statistical Association, 2012, 107: 592-606. DOI URL
[116]	EIDENBERGER J, NEUDORFER B, SIGMUND M, et al. What predicts financial (in)stability? A Bayesian approach[J]. Credit and Capital Markets-Kredit und Kapital, 2017, 50(3): 299-336. DOI URL
[117]	CHEN R B, CHEN Y C, CHU C H, et al. On the determi-nants of the 2008 financial crisis: a Bayesian approach to the selection of groups and variables[J]. Studies in Nonlinear Dynamics & Econometrics, 2017, 21(5): 20160107.
[118]	GUIDOLIN M, HANSEN E, PEDIO M. Cross-asset conta-gion in the financial crisis: a Bayesian time-varying para-meter approach[J]. Journal of Financial Markets, 2019, 45: 83-114. DOI URL
[119]	LANG J H, PELTONEN T A, SARLIN P. A framework for early-warning modeling with an application to banks[J]. Euro-pean Stability Mechanism Working, 2018: 40.
[120]	ALESSI L, BALDUZZI P, SAVONA R. Anatomy of a sove-reign debt crisis: CDS spreads and real-time macroeconomic data[R]. Italy: European Commission-Joint Research Centre, 2019.
[121]	HOLOPAINEN M, SARLIN P. Toward robust early-warning models: a horse race, ensembles and model uncertainty[J]. Quantitative Finance, 2017, 17(12): 1933-1963. DOI URL
[122]	GLASSERMAN P, YOUNG P. Contagion in financial net-works[J]. Office of Financial Research Working Paper, 2015: 15-21.
[123]	MARCUS B, ANDY H, ANNE-CAROLINE H. Comparing minds and machines: implications for financial stability[J]. Oxford Review of Economic Policy, 2021, 37(3): 479-508. DOI URL
[124]	SCHULDENZUCKER S, SEUKEN S, BATTISTON S. De-fault ambiguity: credit default swaps create new systemic risks in financial networks[J]. Management Science, 2020, 66(5): 1981-1998. DOI URL
[125]	GALAT G, MOESSNER R. Macro prudential policy: a litera-ture review[J]. Journal of Economic Surveys, 2013, 27(5): 846-878.
[126]	KARA G I. Systemic risk, international regulation, and the limits of coordination[J]. Journal of International Economics, 2016, 99(1): 192-222. DOI URL
[127]	POSNER E, WEYL E G. Benefit cost analysis for financial regulation[J]. The American Economic Review, 2013, 103(3): 393-397. DOI URL
[128]	BOSMA J J. Dueling policies: why systemic risk taxation can fail[J]. European Economic Review, 2016, 87(1): 132-147. DOI URL
[129]	CLARK E, JOKUNG O. The role of regulatory credibility in effective bank regulation[J]. Journal of Banking & Fin-ance, 2015, 50(1): 506-513.
[130]	CAO J, ILLING G. Regulation of systemic liquidity risk[J]. Financial Markets and Portfolio Management, 2010, 24(1): 31-48. DOI URL
[131]	NUCERA F, SCHWAAB B, KOOPMAN S J. The informa-tion in systemic risk rankings[J]. Journal of Empirical Fin-ance, 2016, 38: 461-475.
[132]	李拉亚. 经济学计算机化研究进展[J]. 经济学动态, 2014(1): 130-137.
	LI L Y. The progress of research on the computerization of economics[J]. Economic Perspectives, 2014(1): 130-137.

分类	方法	文献算法	优点	局限	传统场景
信用风险	监督学习	Kim等人^[21]、Kim等人^[22]、Barboza等人^[23]、Ghodselahi^[24]、Hu等人^[25]、Wang等人^[26]、Ampountolas等人^[27]、Harding等人^[28]	预测精确度优于传统计量技术	未引入非金融变量、数据集规模	预测破产或信用评分
信用风险	无监督学习	Lim等人^[29]、Kou等人^[30]	分类精度优于传统计量技术	聚类算法、数据集规模	识别破产或信用违约风险
市场风险	波动率机器学习	Luong等人^[32]、Park等人^[33]、Luo等人^[34]、LSTM^[35]、Ramos-Pérez等人^[36]、Nizer等人^[39]、Manela等人^[40]	市场波动率优于传统计量方法	实验数据集、文本有效性	市场风险防控
市场风险	组合的机器学习	Papadimitriou等人^[44]、Ban等人^[45]、Pinelis等人^[46]、Li等人^[47]、Li等人^[48]、Jiang等人^[49]、RRL^[50]	组合风险优于传统计量方法	训练测试集规模、模型泛化	风险最小化投资组合
操作风险	监督学习	Bahnsen等人^[53]、Abbasi等人^[54]、Kusaya等人^[55]	欺诈异常检测优于传统统计方法	数据集、模型基础准确性	预防异常欺诈监督学习
操作风险	无监督学习	Nian等人^[57]、Chalapathy等人^[59]、Fiore等人^[60]	异常欺诈检测优于传统统计方法	异常、欺诈检测系统	预防异常欺诈无监督学习
保险风险	索赔的机器学习	Verbelen等人^[63]、Tsai等人^[84]、Ghaffar等人^[65]、Ledig等人^[66]、Li等人^[67]	增强了图像数据集	图像去噪、图像去雾的增强技术	保单索赔
保险风险	死亡的机器学习	Lee等人^[68]、Richman等人^[70]、Hainaut^[71]、Karthikeyan等人^[72]、Sriram等人^[73]、Gourdeau等人^[74]	图像数据集的安全性、实时性	图像数据集数据规模和密度	死亡保险

分类	方法	文献算法	优点	局限	传统场景
信用风险	监督学习	Kim等人^[21]、Kim等人^[22]、Barboza等人^[23]、Ghodselahi^[24]、Hu等人^[25]、Wang等人^[26]、Ampountolas等人^[27]、Harding等人^[28]	预测精确度优于传统计量技术	未引入非金融变量、数据集规模	预测破产或信用评分
信用风险	无监督学习	Lim等人^[29]、Kou等人^[30]	分类精度优于传统计量技术	聚类算法、数据集规模	识别破产或信用违约风险
市场风险	波动率机器学习	Luong等人^[32]、Park等人^[33]、Luo等人^[34]、LSTM^[35]、Ramos-Pérez等人^[36]、Nizer等人^[39]、Manela等人^[40]	市场波动率优于传统计量方法	实验数据集、文本有效性	市场风险防控
市场风险	组合的机器学习	Papadimitriou等人^[44]、Ban等人^[45]、Pinelis等人^[46]、Li等人^[47]、Li等人^[48]、Jiang等人^[49]、RRL^[50]	组合风险优于传统计量方法	训练测试集规模、模型泛化	风险最小化投资组合
操作风险	监督学习	Bahnsen等人^[53]、Abbasi等人^[54]、Kusaya等人^[55]	欺诈异常检测优于传统统计方法	数据集、模型基础准确性	预防异常欺诈监督学习
操作风险	无监督学习	Nian等人^[57]、Chalapathy等人^[59]、Fiore等人^[60]	异常欺诈检测优于传统统计方法	异常、欺诈检测系统	预防异常欺诈无监督学习
保险风险	索赔的机器学习	Verbelen等人^[63]、Tsai等人^[84]、Ghaffar等人^[65]、Ledig等人^[66]、Li等人^[67]	增强了图像数据集	图像去噪、图像去雾的增强技术	保单索赔
保险风险	死亡的机器学习	Lee等人^[68]、Richman等人^[70]、Hainaut^[71]、Karthikeyan等人^[72]、Sriram等人^[73]、Gourdeau等人^[74]	图像数据集的安全性、实时性	图像数据集数据规模和密度	死亡保险

文献算法	文献模型局限性
Kim等人^[21]	限于多类分类支持向量机模型最有效、不同软件平台实验假设
Kim等人^[22]	受限于未引入非金融变量、公司信用评分、融合性集成算法
Barboza等人^[23]	限于交叉验证过拟合、未引入宏观经济数据
Ghodselahi等人^[24]	局限于数据库规模、集成模型融合度
Hu等人^[25]	限于ML算法何时使用、变量选择和数据缺失、最佳拟合优度问题
Wang等人^[26]	受限于工商银行收集公司的数据规模、混合模型集成度
Ampountolas 等人^[27]	限于小额贷款机构的真实数据、特征选择不适用于其他国家和其他机构
Harding等人^[28]	受限于违约数据规模、商业贷款违约处置
Lim等人^[29]	限于小规模数据集、分割数据缩短观测期限
Kou等人^[30]	受限于多准则决策方法与聚类算法不一致、数据规模

文献算法	文献模型局限性
Kim等人^[21]	限于多类分类支持向量机模型最有效、不同软件平台实验假设
Kim等人^[22]	受限于未引入非金融变量、公司信用评分、融合性集成算法
Barboza等人^[23]	限于交叉验证过拟合、未引入宏观经济数据
Ghodselahi等人^[24]	局限于数据库规模、集成模型融合度
Hu等人^[25]	限于ML算法何时使用、变量选择和数据缺失、最佳拟合优度问题
Wang等人^[26]	受限于工商银行收集公司的数据规模、混合模型集成度
Ampountolas 等人^[27]	限于小额贷款机构的真实数据、特征选择不适用于其他国家和其他机构
Harding等人^[28]	受限于违约数据规模、商业贷款违约处置
Lim等人^[29]	限于小规模数据集、分割数据缩短观测期限
Kou等人^[30]	受限于多准则决策方法与聚类算法不一致、数据规模

文献算法	文献模型局限性
Luong等人^[32]	限于高频数据集、Boostrop自助抽样
Park等人^[33]	受限于实验数据规模、数据稀疏、变量数据的高维度
Luo等人^[34]	限于整个样板集、子样本划分及规模
Hochreiter等人^[35]	局限于大样本数据、网络收敛费时
Ramos-Pérez等人^[36]	限于S&P数据集、混合模型融合度
Nizer等人^[39]	受限于公司新闻敏感性、分类法效果
Manela等人^[40]	限于华尔街杂志文本集合、文本的有效性

机器学习解构区域金融风险防控研究进展

Progress on Machine Learning for Regional Financial Risk Prevention

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 21

参考文献 132

相关文章 15

编辑推荐

Metrics

文献算法	文献模型局限性
Papadimitriou等人^[44]	限于标准普尔500指数极端变动、数据过拟合
Ban等人^[45]	受限于KF网站3个数据集、拟合时间花费较长
Pinelis等人^[46]	限于KF等网站月度数据、重复抽样技术
Li等人^[47]	局限于单期均值回归存在性、模型实际泛化性、破产资产
Li等人^[48]	限于数据挖掘与机器学习的融合
Jiang等人^[49]	受限于比特币零滑点、零市场影响假设的设计
Almahdi等人^[50]	限于目标函数收敛性、周收盘价训练集及测试集规模

文献算法	文献模型局限性
Bahnsen等人^[53]	限于信用卡公司数据集、获取交易特征的时间太长
Abbasi等人^[54]	受限于其6项假设、年度或季度文本特征信息集规模
Kusaya等人^[55]	限于财务报表数据准确性、模型基准的准确性
Nian等人^[57]	局限于分类或有序数据集特征、全局异常检测的单聚类策略
Chalapathy 等人^[59]	限于ML各类模型解决异常检测的假设、优点和缺陷
Fiore等人^[60]	受限于信用卡训练数据集、识别欺诈检测系统

文献算法	文献模型局限性
Verbelen等人^[63]	忽略传统的风险因素,未考虑远程信息处理预测变量的成分结构
Tsai等人^[64]	图像无动态模糊和观测噪声,不考虑光学系统的点扩散函数
Ghaffar等人^[65]	图像增强技术的图像去噪和图像去雾时效果不明显
Ledig等人^[66]	高频数据的图像效果不佳,训练时间较长,图像纹理重建有待提高
Li等人^[67]	物联网提取图像先验信息不足,数据传输质量不高,场景没有充分开发和相互补充

文献算法	文献模型局限性
Lee等人^[68]	置信带较窄,死亡风险低估,时间短时低估更严重
Richman等人^[70]	精算和人口统计中神经网络模型预测的不确定性
Hainaut^[71]	未计入传统风险指标因素、舍去远程信息系统预测变量的成分结构
Karthikeyan等人^[72]	图像数据集的数据规模和密度
Sriram等人^[73]	放射科医生对死亡率的概率评分、图像特征信息发现与临床实际表现的差异
Gourdeau等人^[74]	图像数据集规模偏差,治疗临床信息缺失,预测放射轨迹排除“稳定”图像

文献算法	文献模型局限性
Amini等人^[81]	限于弹性测度和资本比例、与Basel协议不一致的最低资本要求、节点特征的观测性
Giudici等人^[84]	受限于金融机构相互关联性、BIS本地银行业数据库规模
Yu等人^[85]	限于控制的资本储备比例、银行发生巨额资产违约
Battiston等人^[88]	限于金融网络复杂性与降低系统风险的此消彼长关联性
Poledna等人^[89]	限于银行网络层数、违约后可回收比率
Grassia等人^[90]	受限于最小拆解单元集规模、非确定性多项式计算难题

文献算法	文献模型局限性
Sarlin^[92]	限于4维数据的复杂性、不同数据实证识别与测算的差异
Sarlin^[93]	受限于宏观审慎数据结构、系统风险可视分析平台
Cerchiello等人^[94]	限于意大利上市银行数量、金融推文大数据规模
Cerchiello等人^[91]	局限于高斯图模型泛化性、金融市场数据与图文数据的融合性
Flood等人^[95]	限于数据质量、数据标准和数据整合、元数据管理
Jagadish等人^[96]	受限于大数据的时效性、不完整性、私密性及数据整合加总技术
Nyman等人^[99]	限于市场文本信息集、噪音与信号的识别测度
Chiang等人^[100]	局限于台湾省电子产业情绪数据、监督学习与半监督学习的融合性
Meyer等人^[102]	受限于金融新闻机构和网站新闻资源、自然语言不规范、离散的用词向量空间

文献算法	文献模型局限性
Duttagupta等人^[104]	限于金融稳健与宏观基本面复杂交互性、央行独立性及公司治理脆弱时间序列数据缺失
Ward^[108]	受限于分类树合成集合与多种预测变量的融合
Casabianca等人^[109]	限于银行和宏观经济指标集成数据集、宏观经济累计失衡
Fan等人^[115]	局限于风险协方差矩阵设计、因子模型准确率的未知性
Guidolin等人^[118]	限于ABS债券、国债、公司债券和股票数据集的规模和结构
Lang等人^[119]	受限于政策偏好预期、预测期限、欧盟银行数据集、样本预选择
Battiston等人^[75]	限于金融经济与生态病理等学科交叉发展,以及数据、方法、指标等复杂融合
Marcus等人^[123]	局限于AI如何实现金融稳定、承担监管责任、设计经济政策、进行自动监管
Schuldenzucker 等人^[124]	受限于信用违约互换合约估值的一致性向量、加权依赖关系图表示的高度非线性

文献算法	文献模型局限性
Kara^[126]	限于不对称国家资本充足要求的协调限制、国家宏观审慎监管不足
Posner等人^[127]	受限于收益成本方法的稳健经济基础、健全金融监管体系
Bosma^[128]	限于政策工具之间相互作用、建设性模糊假设
Clark等人^[129]	局限于随机干预成本、不成熟干预时机

类别	方法	文献算法	优势	缺陷	适应场景
金融系统风险	金融网络风险	Amini等人^[81]、Giudici等人^[84]、Yu等人^[85]	风险敞口传导测度优于传统方法	受限于金融机构关联性、违约资产	金融网络的风险敞口与传导
	金融网络风险	Battiston等人^[88]、Poledna等人^[89]、Grassia等人^[90]	捕捉网络结构的关键特征	受限于网络多层结构的复杂度	金融机构网络的结构
	大数据分析	Sarlin^[92]、Sarlin^[93]、Cerchiello等人^[94]	大数据分析优于传统计量方法	限于实验、推文大数据规模	系统风险大数据分析
		Cerchiello等人^[91]、Flood等人^[95]、Jagadish等人^[96]	增强丰富了传统数据	限于数据质量、时效性、融合性	系统风险的数据问题
		Nyman等人^[99]、Chiang等人^[100]、Meyer等人^[102]	情绪分析预测系统风险	限于用词向量空间推文新闻数据	金融市场的情绪分析
	金融稳定性	Duttagupta等人^[104]、Ward^[108]、Casabianca等人^[109]	风险预警优于传统计量模型	受限于宏观指标数据、决策树集成	适于金融稳定的决策树
		Fan等人^[115]、Guidolin等人^[118]、Lang等人^[119]	建立稳定稀疏投资组合	受限于数据选择、规模、结构	适于金融稳定的稀疏模型
		Battiston等人^[75]、Marcus等人^[123]、Schuldenzucker等人^[124]	金融稳定的网络方法	受限于多学科数据、方法融合	适于金融稳定的无监督学习
	风险监管量化	Kara^[126]、Posner等人^[127]、Bosma^[128]、Clark等人^[129]	系统风险监管的政策依赖、优化网络分析	限于监管网络、监管成本、金融创新等	风险监管

[1]	武美, 丁怡彤, 赵建立. 改进的增量式动静结合协同过滤方法[J]. 计算机科学与探索, 2022, 16(9): 2089-2095.
[2]	刘学文, 王继奎, 杨正国, 李冰, 聂飞平. 密度峰值隶属度优化的半监督Self-Training算法[J]. 计算机科学与探索, 2022, 16(9): 2078-2088.
[3]	罗逸轩, 刘建华, 胡任远, 张冬阳, 卜冠南. 融合经验共享Q学习的粒子群优化算法[J]. 计算机科学与探索, 2022, 16(9): 2151-2162.
[4]	朱炳宇, 刘朕, 张景祥. 融合Grad-CAM和卷积神经网络的COVID-19检测算法[J]. 计算机科学与探索, 2022, 16(9): 2108-2120.
[5]	任宁, 付岩, 吴艳霞, 梁鹏举, 韩希. 深度学习应用于目标检测中失衡问题研究综述[J]. 计算机科学与探索, 2022, 16(9): 1933-1953.
[6]	萨日娜, 李艳玲, 林民. 知识图谱推理问答研究综述[J]. 计算机科学与探索, 2022, 16(8): 1727-1741.
[7]	谢娟英, 张凯云. XR-MSF-Unet：新冠肺炎肺部CT图像自动分割模型[J]. 计算机科学与探索, 2022, 16(8): 1850-1864.
[8]	陈洋, 王士同. 多样性正则化极限学习机的集成方法[J]. 计算机科学与探索, 2022, 16(8): 1819-1928.
[9]	杨政, 邓赵红, 罗晓清, 顾鑫, 王士同. 利用ELM-AE和迁移表征学习构建的目标跟踪系统[J]. 计算机科学与探索, 2022, 16(7): 1633-1648.
[10]	韩毅, 乔林波, 李东升, 廖湘科. 知识增强型预训练语言模型综述[J]. 计算机科学与探索, 2022, 16(7): 1439-1461.
[11]	张壮, 王士同. 不平衡数据的Takagi-Sugeno-Kang模糊分类集成模型[J]. 计算机科学与探索, 2022, 16(6): 1374-1382.
[12]	赵家波, 游晓明, 刘升. 结合价格波动策略与动态回溯机制的蚁群算法[J]. 计算机科学与探索, 2022, 16(6): 1390-1404.
[13]	谢欣彤, 胡悦阳, 刘譞哲, 赵耀帅, 姜海鸥. 传播用户代表性特征学习的谣言检测方法[J]. 计算机科学与探索, 2022, 16(6): 1334-1342.
[14]	刘雅芬, 郑艺峰, 江铃燚, 李国和, 张文杰. 深度半监督学习中伪标签方法综述[J]. 计算机科学与探索, 2022, 16(6): 1279-1290.
[15]	董文波, 孙仕亮, 殷敏智. 医学知识推理研究现状与发展[J]. 计算机科学与探索, 2022, 16(6): 1193-1213.