IJDSBDA International Journal of Data Science and Big Data Analytics 2710-2599 SvedbergOpen ijdsbda-1-1-001 /10.51483/IJDSBDA.1.1.2021.1-19 Review Article Algorithms in future insurance markets ŚmietankaMałgorzata 1 * KoshiyamaAdriano 2 TreleavenPhilip 3 1University College London, Gower St, Bloomsbury, London WC1E 6BT, United Kingdom. E-mail: malgorzata.wasiewicz.17@ucl.ac.uk 2University College London, Gower St, Bloomsbury, London WC1E 6BT, United Kingdom. E-mail: adriano.koshiyama.15@ucl.ac.uk 3University College London, Gower St, Bloomsbury, London WC1E 6BT, United Kingdom. E-mail: p.treleaven@ucl.ac.uk *Corresponding author: Małgorzata Śmietanka, University College London, Gower St, Bloomsbury, London WCĴE 6BT, United Kingdom. E-mail: malgorzata.wasiewicz.17@ucl.ac.uk 02 2021 1 1 1 19 Abstract

This paper reviews the impact of data science and artificial intelligence (AI) on future ‘data- driven’ insurance markets. The impact of insurance automation (driven by so-called Black Swan1 events such as Covid-19) mirrors the impact of algorithmic trading that changed radically the capital markets (Koshiyama et al., 2020). The data science technologies driving change include: Big data, AI analytics, Internet of Things, and Blockchain technologies. These technologies are important since they underpin the automation of the insurance markets and risk analysis, and provide the context for the algorithms, such as AI machine learning and computational statistics, which provide powerful analytics capabilities. New AI algorithms are constantly emerging, with each ‘strain’ mimicking a new form of human learning, reasoning, knowledge, and decision-making. The current main disrupting forms of learning include deep learning, adversarial learning, federated learning, transfer and meta learning. Albeit these modes of learning have been in the AI/ML field more than a decade, they are now more applicable due to the availability of data, computing power and infrastructure. These forms of learning have produced new models (e.g., long short-term memory, generative adversarial networks) and leverage important applications (e.g., Natural Language Processing, Adversarial Examples, Deep Fakes, etc.). These new models and applications will drive changes in future insurance markets, so it is important to understand their computational strengths and weaknesses. The contribution of this paper is to review the data science technologies and specifically AI algorithms, their computational strengths and weaknesses, and discuss their future impact on the insurance markets.

 Keywords Insurance AI Machine Learning Algorithms review References AdewumiA.O.AkinyeluA.A. (2017). A survey of machine-learning and nature-inspired based credit card fraud detection techniques. International Journal of System Assurance Engineering and Management. 8, 937–953. AllanN.CantleN.GodfreyP.YinY. (2012). A review of the use of complex systems applied to risk appetite and emerging risks in ERM practice: Recommendations for practical tools to help risk professionals tackle the problems of risk appetite and emerging risk.. British Actuarial Journal, 18(1), 163–234. ArlotS.CalisseA. (2009). A survey of cross-validation procedures for model selection. arXiv:0907.4728. ArulkumaranK.DeisenrothM.P.BrundageM.BharathA.A. (2017). Deep reinforcement learning: a brief survey. IEEE, 34(6), 26–38. BachS. et al. (2015). On pixel-wise explanations for non-linear classifier decisions by layer-wise relevance propagation. PLOS ONE. BatrincaB.TreleavenP.C. (2015). Social media analytics: a survey of techniques, tools and platforms. AI and Society, 30, 89–116. BäumerH. P. (2000). Statistical analysis of extreme values - from insurance, finance, hydrology and other fields. Computational Statistics, 5, 307–311. BrockA.DonahueJ.SimonyanK. (2018). Large scale gan training for high fidelity natural image synthesis. arXiv:1809.11096. BrownleeJ. (2011). Clever Algorithms: Nature-Inspired Programming Recipes. s.l.:Lulu.com CambriaE.SchullerB.LiuB. (2013). Statistical approaches to concept-level sentiment analysis. Intelligent Systems, IEEE, 28(3), 6–9. CarliniN.AthalyeA.PapernotN.BrendelW.RauberJ.TsiprasD.GoodfellowI.MadryA.KurakinA. (2019). On evaluating adversarial robustness. arXiv:1902.06705 CaronM.BojanowskiP.JoulinA.DouzeM. (2018). Deep clustering for unsupervised learning of visual features.. The European Conference on Computer Vision (ECCV), pp. 132–149. ChenS.MaK.ZhengY. (2019). Transfer learning for 3d medical image analysis. arXiv preprint arXiv:1904.00625 ChenX.DuanY.HouthooftR.SchulmanJ.SutskeverI.AbbeelP. (2016). InfoGAN: Interpretable representation learning by information maximizing generative adversarial nets.. arXiv:1606.03657. CholletF. (2017). Deep Learning with Python. s.l.:Manning. CreswellA.WhiteT.DumoulinV.ArulkumaranK.Biswa SenguptaB.BharathA.A. (2018). Generative adversarial networks: An overview. IEEE Signal Processing Magazine, 35(1), 53–65. DaiW.XueG. R.YangQ.YuY. (2007). Co-clustering based classification for out-of-domain documents. KDD ‘07: Proceedings of the 13 th ACM SIGKDD international conference on Knowledge discovery and data mining, pp. 210 – 219. DenevA.AmenS. (2020). The Book of Alternative Data: A Guide for Investors, Traders and Risk. s.l.:Wiley. DevlinJ.ChangM.W.LeeK.ToutanovaK. (2018). BERT: Pre-training of deep bidirectional transformers for language understanding. arXiv:1810.04805. DiaoL.WangN. (2019). Research on premium income prediction based on LSTM neural network. Advances in Social Sciences Research Journal, 6(11), 256–260. DíazZ.SegoviaM.J.FernándezJ.del PozoE.M. (2005). Machine learning and statistical techniques. An application to the prediction of insolvency in spanish non-life insurance companies.. The International Journal of Digital Accounting Research, 5(9), 1–45. EfronB.HastieT. (2016). Computer Age Statistical Inference: Algorithms, Evidence and Data Science. Cambridge University Press, 5. Fei-FeiL.FergusR.PeronaP. (2003). A bayesian approach to unsupervised one-shot learning of object categories. In Proceedings Ninth IEEE International Conference on Computer Vision, 1134–1141. GiarratanoJ.C.RileyG. (1998). Expert Systems. s.l.:PWS Publishig Co.. GleaveA.DennisM.WildC.KantN.LevineS.RussellS. (2019). Adversarial policies: attacking deep reinforcement learning.. arXiv:1905.1 0615. GoodfellowI.BengioY.CourvilleA. (2016). Deep Learning. s.l.:MIT Press book. GoodfellowI.J.Pouget-AbadieJ.MirzaM.XuB.Warde-FarleyD.OzairS.CourvilleA.BengioY. (2014). Generative Adversarial Networks. arXiv:1406.2661. GravesA. (2014). Generating sequences with recurrent neural networks. arXiv:1308.0850. GravesA.JaitlyN. (2014). Towards end-to-end speech recognition with recurrent neural networks. Proceedings of the 31st International Conference on Machine Learning, 32(2), 1764–1772. GravesA.LiwickiM.FernándezS.BertolamiR.BunkeH.SchmidhuberJ. (2009). A novel connectionist system for unconstrained handwriting recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence, 31(5), 855–868. GravesA.MohamedA.R.HintonG. (2013). Speech recognition with deep recurrent neural networks. arXiv:1303.5778. HastieT.TibshiraniR.FriedmanJ. (2009). The elements of statistical learning: data mining, inference, and prediction. Second ed. s.l.:Springer. HigginsI.LiwickiM.FernándezS.BertolamiR.BunkeH.SchmidhuberJ. (2017). DARLA: Improving zero-shot transfer in reinforcement learning. arXiv:1707.08475. HochreiterS.SchmidhuberJ. (2006). Long short-term memory. Neural Computation, 9(8), 1735–1780. HuangL.JosephA.D.NelsonB.RubinsteinB.I.P.TygarJ.D. (2011). Adversarial machine learning. AISec ‘11: Proceedings of the 4th ACM workshop on Security and artificial intelligence, pp. 43–58. IlkerK.MehmetG.KemalK. (2015). An interactive machine-learning-based electronic fraud and abuse detection system in healthcare insurance.. Applied Soft Computing, 36, 283–299. JuszczakP.AdamsN.M.DavidJ.HandD.J.WhitrowC.WestonaD.J. (2008). Off-the-peg and bespoke classifiers for fraud detection. Computational Statistics and Data Analysis, 52(9), 4521–4532. KarpathyA.JohnsonJ.Fei-FeiL. (2015). Visualizing and understanding recurrent networks. arXiv:1506.02078v2. KeneshlooY.RamakrishnanN.ReddyC. K. (2019). Deep transfer reinforcement learning for text summarization. arXiv:1810.06667. KimB.WattenbergM.GilmerJ.CaiC.WexlerJ.ViegasF.SayresR. (2018). Interpretability beyond feature attribution: quantitative testing with concept activation vectors (TCAV). Proceedings of the 35th International Conference on Machine Learning, pp. 2668–2677. KirosR.SalakhutdinovR.ZemelR.S. (2014). Unifying visual-semantic embeddings with multimodal neural language models. arXiv:1411.2539. KolchynaO.SouzaT.P.TreleavenP. (2015). Twitter sentiment analysis. arXiv:1507.00955. KoshiyamaA.FiroozyeN. (2019). Avoiding backtesting overfitting by covariance-penalties. The Journal of Financial Data Science. KoshiyamaA.FiroozyeN.TreleavenP. (2020). Algorithms in future capital markets. SSRN: https://papers.ssrn.com/ sol3/papers.cfm?abstract_id=3527511 KuoK. (2019). Generative synthesis of insurance datasets. arXiv:1912.02423. KurakinA.GoodfellowI.BengioS. (2016). Adversarial machine learning at scale. arXiv:1611.01236. LakkarajuH.KamarE.CaurunaR.LeskovecJ. (2017). Interpretable and explorable approximations of black box models. arXiv:1707.01154. LiQ. et al. (2019). A survey on federated learning systems: vision, hype and reality for data privacy and protection. arXiv:1907.09693. LundbergS.Su-InL. (2017). A unified approach to interpreting model predictions. arXiv:1705.07874. MakhzaniA. et al. (2016). Adversarial autoencoders. arXiv:1511.05644. MaranoP.NoussiaK. (2019). Insurtech: a legal and regulatory view. In: Smart Contracts in Insurance: A Law and Futurology Perspective. s.l.:Springer, pp. 101–125. MayR.M.LevinS.A.SugiharaG. (2008). Complex systems. ecology for bankers. Nature. 451, 893–893. MeyerM. (2014). Continuous integration and its tools. IEEE Software, 31(3), 14–16. MiyatoT.DaiA.M.GoodfellowI. (2017). Adversarial training methods for semi-supervised text classification.. arXiv:1605.07725. MolnarC. (2020). Interpretable machine learning. s.l.:Lulu.com. MontavonG.SamekW.MüllerK.R. (2017). Methods for Interpreting and Understanding Deep Neural Networks. arXiv:1706.07979. NguyenA. et al. (2016). Synthesizing the preferred inputs for neurons in. Advances in Neural Information Processing Systems 29 (NIPS 2016). NicolaeM.I. et al. (2019). Adversarial robustness toolbox. v0.4.0. arXiv:1807.01069. PintoL.DavidsonJ.SukthankarR.GuptaA. (2017). Proceedings of the 34th International Conference on Machine Learning. Volume 70. PoliR.LangdonW.B.McPheeN.F. (2008). A field guide to genetic programming.. s.l.:Lulu.com. QinC. et al. (2019). Verification of non-linear specifications for neural networks. arXiv:1902.09592. RadfordA. et al. (n.d.). Language models are unsupervised multitask learners. OpenAIBlog, 1(8). RainaR.NgA.Y.KollerD. (2006). Constructing informative priors using transfer learning. ICML ‘06: Proceedings of the 23rd international conference on Machine learning, pp. 713–20. RibeiroM.T.SinghS.GuestrinC. (2016). Why should i trust you?: explaining the predictions of any classifier. arXiv:1602.04938. RichmanR. (2018). AI in actuarial science. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3218082 RussellS.NorvigP. (2020). Artificial intelligence: a modern approach. 4th ed. s.l. :Pearson. SalehK.HossnyM.NahavandiS. (2017). Driving behavior classification based on sensor data fusion using LSTM recurrent neural networks. 2017 IEEE 20th International Conference on Intelligent Transportation Systems (ITSC). SalehK.HossnyM.NahavandiS. (2017). Driving behavior classification based on sensor data fusion using LSTM recurrent neural networks. 2017 IEEE 20th International Conference on Intelligent Transportation Systems (ITSC). SiddhantA.GoyalA.MetallinouA. (2019). Unsupervised transfer learning for spoken language understanding in intelligent agents. In Proceedings of the AAAI Conference on Artificial Intelligence, Volume 4959-4966. SutskeverI.VinyalsO.LeQ. V. (2014). Sequence to sequence learning with neural networks. arXiv:1409.3215. SuttonR.S.BartoA.G. (2018). Reinforcement learning: An introduction. s.l.:MIT Press. SzegedyC. et al. (2013). Intriguing properties of neural networks. arXiv:1312.6199. TaylorS. J. (2014). Agent-Based Modeling and Simulation. s.l.:Springer. TreleavenP.BarnettJ.KoshiyamaA. (2019). Algorithms: Law and regulation. Computer, 52(2), 32–40. TreleavenP.GalasM.LalchandV. (2013). Algorithmic trading review. Communications of the ACM, 56(11), 76–85. van LiebergenB. (2017). Machine learning: A revolution in risk management and compliance?. Journal of Financial Transformation, 45, 60–67. VinyalsO. et al. (2015). Grammar as a foreign language. arXiv:1412.7449. WachterS.MittelstadtB.RussellC. (2018). Counterfactual explanations without opening the black box: Automated Decisions and the GDPR. Harvard Journal of Law and Technology, 31(2). WeishanD. et al. (2016). Characterizing driving styles with deep learning. arXiv:1607.03611. WoodS. N. (2017). Generalized additive models. s.l.:Chapman and Hall/CRC. WüthrichM. V. (2018). Machine learning in individual claims reserving.. Scandinavian Actuarial Journal, 465–480. XuJ.ChenH. (2005). Criminal network analysis and visualization. Communications of the ACM, 48(6), 100–107. XuK. et al. (2016). Show, Attend and Tell: Neural Image Caption Generation with Visual Attention. arXiv:1502.03044 YangQ. et al. (2019). Federated learning. synthesis lectures on artificial intelligence and machine learning. s.l.:Morgan and Claypool Publishers. YangW.KoshiyamaA.S. (2019). Assessing qualitative similarities between financial reporting frameworks using visualization and rules: COREP vs. pillar 3.. Intelligent Systems in Accounting, Finance and Management, 26(1), 16 – 31. YaqiL.ChunY.WeiL.MaozhenL. (2018). principle component analysis-based random forest with the potential nearest neighbor method for automobile insurance fraud identification. Applied Soft Computing, 70(September 2018), 1000–1009. YosinskiJ. et al. (2015). Understanding neural networks through deep visualization. arXiv:1506.06579. ZintgrafL.M.ChohenT.S.AdelT.WellingM. (2017). Visualizing deep neural network decisions: prediction difference analysis. arXiv:1702.04595.