bringing machine learning to regulatory compliance · 2019-11-11 · bringing machine learning to...
TRANSCRIPT
Bringing Machine Learning to Regulatory Compliance
– Gwee Yee Teong Sagar Sinha– Head of FSI Sales, APAC VP Sales & Channels, APAC & ME
– HPE Tookitaki
– September 19th, 2019
Why should you be interested in AI / ML / DL?
2
Everyone wants AI / ML / DL and advanced analytics….
AI and advanced analyticsinfrastructure could constitute
15-20% of the market by 20211
AI and advanced analytics represent2 of top 3 CIO priorities
Enterprise AI adoption
2.7X growth in last 4 years2
….but face many challenges
Use cases
New roles, skill gaps
Culture and change
Data preparation
Legacy infrastructure
1 IDC. Goldman Sachs. HPE Corporate Strategy.20182 Gartner - “2019 CIO Survey: CIOs Have Awoken to the Importance of AI”
HPE helps “unlock your data with AI”
3
AI-driven OperationsAcross edge to cloud infrastructure
Gain faster insightswith scalable AI solutions & services
Optimize infrastructurewith the latest AI technologies
AI within our products
AI for our customers
AI for the future
SBR Technology Excellence AwardAI for Banking
Accreditation@SGD (2017)Only 21 of 5000 Companies Accredited
QCFintech Accenture
1st Place: MAS Fintech Awards (2016)SME Category
Awards & Accolades
FundedNov 2014 in SingaporeFounded
WEF Technology Pioneer 2019
About Tookitaki
Top 7 AI in FinServ in APAC (2017)
Top AI Startups in Singapore (2017)
• Anti-money laundering suite (AMLS) • Reconciliation suite (RS)
Enable sustainable compliance programmes at financial institutions
Products
Vision
Values
• Prime Focus on Clients• Grit towards pursuit of excellence• Teamwork and family ambience• Continuous innovation
Company
AdvisoryBoard
• HQ in Singapore; Offices in India & US• 100% R&D and Product Development in Singapore• Core Team with 30+y in Finance, Machine Learning & Big Data
• Senior Advisor with 25+ yrs Retail & Weath (HSBC, Citi, Barclays)• Jungle Ventures• McKinsey & Co
About Tookitaki
Money Laundering is a continuing challenge
8
Soundness of the banking system is challenged
Banks are still using multiple compliance technologies, and employing resources in separate teams for monitoring, screening and reporting
Escalating costs pose a strategic risk
laundered annually growing at 10+%
fines paid by banks in the US alone since the financial crisis of 2008 until
2017
$321 billion
Banking and Financial Services (BFS) industry spent has gone up by 20+% for compliance
LexisNexis, BCG, Bloomberg, UN
Money Laundering is a continuing challenge
$1.8 trillion
Money Laundering is a continuing challenge
9
Compliance spend to touch $118 billion
globally by 2020.
Regulatory fines to top ~$400 billion by 2020 just in US and
Europe
Increasing regulatory changes, alerts volume and cost of compliance together with depleting RoI on manpower investment and hefty fines raise question on sustainability of current AML/CFT programs
Annual average compliance spend is common for a +$10-billion bank.
~ 17 million
Compliance spend is on manpower. Size of compliance team has increased 10x in last 5 years.
50%
Why advanced Tech: AI and Machine Learning
10
Gen 0 & 1
2000’s era 2010’s era Current era
Gen 2 Tookitaki (Gen 3)
Rule-based Detection + Watch List match
ML-Driven ML-Driven + Self-Learning +Unknown Detection +Explainable
Throws too many false alerts. Cannot detect new suspicious cases,
and no self learning
Reduce false alerts. Detect new fraud cases with
unsupervised model, refresh by data scientists
Detection of new cases, self-learning and highly explainable models
Tookitaki AMLS solution features & capabilities
11
Tookitaki’s ML-powered Anti Money Laundering Suite (AMLS)
Smart rules creation
Features
Detect unknown unknowns
(Unsupervised ML)
AlertsPrioritization
(Supervised ML)
Improved Matching and Network Detection
Explanation and Audit Trail
Investigation User Interface
Tookitaki Analyticsplatform
Smart typologies repository
Value drivers
Increase in true positives for monitoring
Alert triaging and reduction in false positives for monitoring and screening
Self learning models
Dynamic Segmentation
(ML)
Tookitaki AMLS value proposition
12
Improved Network Detection & Dynamic
Segmentation
Explanation and Audit Trail
Prioritization
Scenario Explainability
Investigator Insights
Detect Unknown Unknowns
Reduces Cost of AML
Operations
Reduces Enterprise AML Risk Accurate
Predictions
1
4
SeamlessIntegration
2
Full Explainability & Auditability
3
Self-Learning
Solutioned Uniquely
Quantifiable business value and impact
13
50% Reduction in False Positive
Hits (Corporate
Names)
60% Reduction in False Positive
Hits(Individual
Names)
After Deploying Tookitaki AML Name/Client Screening
Identified 12 Unknown Cases –
9 Filed as STRs
5% Increase in True Positive
Alerts
40% Reduction in False Positive
Alerts
After Deploying Tookitaki AML Transaction Monitoring
Result in High Risk: • Huge Backlogs• Alert Ageing • Investigator churn• Missed activity
Nearly All Alerts Are False Positives (93-97%)
Massive Numbers of Alerts for Acceptable Coverage
Prior to Implementing Tookitaki: Traditional Systems Fall Well Short
Millions (USD) Saved Substantially Reduced
Compliance Costs Significantly Reduced Risk
Tookitaki’s competitive advantage
14
Competitive Advantage
ML Models & User
Interactions Are
AuditableSelf-
Learning & Continuous
Model Improvement
Smart Typology
Framework to Detect Unknown
CasesDynamic Typology
Repository to Evolve Detection Strategies
Intelligent Triaging to
Bucket Alerts in Relevant
Risk Bands
Fully Explainable
A Sustainable Compliance Program for AML
Revolutionize Regulatory Compliance
Innovating Beyond Rules-Based AML
HPE GreenLake Big Data-as-a-ServiceFaster time to business insights
16
Per Nodes
DesignImplementOperate
SymmetricAsymmetric
Best fit for Hadoop and AMLSReference Architecture on HPE x86 platform
Operated for youFree your teams for more valuable contribution
Design and ConsultingHigh performance and high availability platform
Tailored components to fit your needsEnterprise-grade Cloudera
HPE AI and Data Platform solutions
17
AI Layer -Platform Focus
Data Platform Layer
Data Deployment Layer
AI SolutionsLayer
Advisory , professional and operational services Custom Use Case / ISV Solutions
Machine Learning as a service Automated AI Tools
Hadoop Ecosystem Fast Data &
Edge Data Cloudera Data Flow
Big Data
AI and BigData Deployments
HPE Apollo 6500 HPE Apollo 4200optimized for Big Data Analytics and other data storage intensive workloads
The enterprise bridge to accelerated computing
HPE Apollo 2000The bridge to enterprise scale-out architecture
HPE DL 380unprecedented levels of performance for databases and analytic workloads
AMLReconciliation
Smart LendingInsurance ClaimsTreasury ComplianceDriverless AI
Wherever you are on your AI journey, HPE can help
18
ExploreHPE Artificial IntelligenceTransformation Workshop
HPE AI/ Deep learning Trainings
Integrate AI to the existing applications
Manage change, train your teams and AI systems
Understand AI, data, analytics outcomes
and challenges
Conduct a Proof-of-Value for select use cases
Explore the best use cases and technologies
Align teams around common
vision & plan
Gather and validate your use cases and data
Modernize data, compute, storage platforms
Protect, manage and consume data
and analytics
ExperimentAI / ML Proof of Value Service
HPE Agile AI Design and Planning Service
EvolveHPE GreenLake
HPE Big Data/ AI Implementation