ping an s ai financial service
play

Ping Ans AI + Financial Service Mei Han Director of Ping An - PowerPoint PPT Presentation

Ping Ans AI + Financial Service Mei Han Director of Ping An Technology, US Research Labs Talk S9863 1 Ping Ans Development and Strategy Corporate Introduction Domestic and Foreign Rising on Fortunes List


  1. Ping An’s AI + Financial Service Mei Han Director of Ping An Technology, US Research Labs Talk � S9863 �������

  2. 1 Ping An’s Development and Strategy Corporate Introduction

  3. Domestic and Foreign Rising on Fortune’s List honors • No.16 in 2017 Forbes Top 2000 companies 2018 No.29 • No.1 in global insurance companies 2017 No.39 • No.61 in 2017 Top 100 global most valuable brands 2016 No.41 • No.1 in global Insurance companies 2015 No.96 2014 No,128 • No.79 in Brand Finance 2017 Top 500 global most valuable brands 2013 No.181 • No.1 in global Insurance companies 2012 No.242 2011 No.328 • China’s most respected company for 16 years 2010 No.383 • Best corporate citizen for 12 times 2008 No.462 • Asian excellence awards, Best CEO, 1988 Officially Established Best CFO for years. 1

  4. One-Day Snapshot of Ping An No.7 in global listed financial companies Clients Market size No.1 in global Insurance companies 150 million clients one Ping An client for every 10 Chinese 2.6 billion RMB per day Revenue 970 billion RMB per year Research Expenditure 270 million RMB per day 1% of revenue Net Profit almost 90 billion RMB per year over 1 billion RMB in 2017 300 million RMB per day Taxes Insurance Cases almost 100 billions in 2017 31,000 cases per day 384 million RMB per day 98.7% cases compensated in one day Insurance Claims almost 140 billion RMB per year Call Center 1.8 million internal and external employees Employees 96 thousand calls per day one Ping An-er in every 800 Chinese 350 million calls per year over 22 thousand IT and research experts 2

  5. Strategic Objectives World-leading technology-powered personal financial services provider Industry Focuses Pan Financial Assets Pan Health Care Development Models Finance + Technology Finance + Ecosystem Health Real Asset Auto Smart Financial Care Estate N Pillars Insurance Banking Cities Services Services Services Finance Management Ecosystem Ecosystem Ecosystem Ecosystem Ecosystem Tech platform Group Technology Platform AI - Block Chain – Cloud - Big Data - Security 3

  6. ���������������������������� ����������������� ������������� ���� ��������������������� ��������������������� ������ ����� ������ ��������������� �������������� ����� ���������� ����������� ���� ����������� ���������� ����������� ��������� ��������� � �� ���������������� ����� ����������� ������������� �������������� ������������������������� ������������������������������������������� ����������� ���� 4

  7. Ping An Advantages on AI + Finance Scenarios & Algorithms Platform Big Data Specialists All Licenses 99.84% 8 integrated Face recognition Data centers finance company accuracy Finance data Deep Learning 30 Years 99%+ Cloud platform Financial industry Voiceprint Healthcare data experiences recognition Million+ accuracy Service data Samples 400+ Parallel computing AI application 400+ speed scenarios Technology patents 5

  8. Ping An Brain AI Engine Risk Control Fraud Detection Precision Marketing Health Management Data management Risk pricing Claims fraud identification Intelligent health assistant platform Credit investigation Network fraud identification Disease prediction LBS analysis Applications Operation Optimization Intelligent Finance Robotics Reports Robot server Service flow optimization Statistical analysis Program trading Voice recognition Salesman training Data visualization Volatility analysis Knowledge graph optimization Unstructured Prediction Detection BI Structured Deep AI Ping An Brain Statistical AI Engine Deep Learning Text Mining Feature Classification Time series Clustering Reinforcement Knowledge extraction/ Topic analysis Rules learning Graph selection Identification Relational Engine Transfer Image Categorical Expert system network Report learning Recognition regression visualization User Channel Product Profiling Basic, wealth, credit, spending, health, … Function, business, cycle, price, profit Types, user behaviors Data Storage Data Processing and Management Cleaning, masking, integration, security, monitoring Compression, Distribution Data Data Acquisition Various data sources 6

  9. 2 Ping An’s Smart Car Insurance Claims Solution

  10. Introduction of Car Insurance Claims Pain points of traditional car insurance claims Ping An’s solution of smart car insurance • Construct Deep Neural Network for vehicle detection and damage • Car insurance claims process is lengthy and complicated classification • The duration from accident occurrence to insurance • Achieve over 90% classification and detection accuracy compensation is long • Finish model training and inference on NVIDIA GPUs • Owner’s claim experience needs to be improved • Use TensorRT for fast network optimizing and inference accelerating The process of Ping An’s smart car insurance solution Loss Intelligent risk Intelligent recognition blocking setting Take Image photos Component Automatic intelligent partitioning pricing processing 7

  11. ��-2��:�����0������������-��������� ��������� ������ ��������� �����-����� X 12 ����������������������� �������� �� ���� ��� ����������2����2�-�������-�������������1�� • -�����0�-������������ ������������������ �0����������������1 �����������������-A�����2�12��2���12����0��� • �2�������-����� �������������� �����������-����������-��-����� • �����-� 8

  12. Car Image Classification Feature extraction Car model classification Actual Case Car type classification Local contour features Local part features Combination feature Whole Car Image BMW 320 (640*360 pixels) 8 fully-connected 30 convolution Claim images from TOYOTA PRIUS layers layers client A Audi A6 Now: 0.15s Past:5s � AI Algorithm � � Human judge � 10 million compensation cases per year Over 40 million vehicle images Over 90% acurracy Over 10 thousand car insurance investigators 38 layers Deep Neural Network 9

  13. Intelligent Vehicle Damage Assessment & Claim • Are the pictures acceptable? Image Enhancement Image Alignment De-Noising Image Editing Detection 10

  14. Intelligent Vehicle Damage Assessment & Claim Rearview mirror(right) Right front leaf • Part Segmentation and Recognition: Part number 13328534 Part number 25886816 Steel ring Right headlight Intelligent Classification Part number 9598038 Part number 13354818 Front bumper Front glass Part number 20827116 Part number 20864731 11

  15. Smart Accessory Model Selection Car Model Library Accessory Library ��(�(�(�)� �������� FAW Audi A6L C7 (2012) ��(�(��)����� ���� ���� ���� ������ ���� �����8���E8B ���������� ���� Image Connect �����8���E8B �1�������� ���� ��� ���50� �D7���3��� ���4� Recognition �B��C��D�A8B ������� ������� ���� ��� ���50� �D7���3��� ���4� �B��C��D�A8B ������������12 ���� ��� ���50� �D7���3��� ���4� �89C��867����C ������ ���� ��� ��� � �50� �D7���3��� ���4� �89C��867����C ����� ����� ���� ��� ���50� �D7���3��� ���4� �89C��B��C� �� ������ � � ��� �8�78B �89C��B��C� ������ ��� � �� �8�78B The Chosen Car �� The Chosen Accessory Model Library Lock Adding VIN Code 12

Recommend


More recommend