Annomaly

Annomaly

Detecting potential fraud, risk & non-compliance

Features

BACKGROUND

In 2016, financial institutions (FIs) filed 2 million SARs (suspicious activity reports) with FinCEN. During 2008—2018, ~90% of US government enforcement actions involved $27B in penalties.

KEY FEATURES

  • Harmonizes originator & beneficiary names, addresses, bank names, etc.
  • Resolves “bleeding” of data across different columns, e.g., name, address
  • Classifies entities – individual, industry, government, not-for-profit, etc.
  • Reduces false positives using Scry’s multidimensional approach of using ensemble of proprietary ML, NLP & knowledge graph algorithms
  • Automates the creation of richer 360o profile of customers

BACKGROUND

Out of ~ 21B ACH transactions worth $46.8T in 2017, ~$1.4B were fraudulent. Financial institutions use business rules to out-sort fraudulent transactions with high false-positive rate due to constant changes in fraudster schemes.

KEY FEATURES

  • Uses 500+ features to generate out-sorts
  • Captures features from transactions, contact centers, user behavior, device connectivity and geolocation
  • Provides “Pareto Frontier” to show trade-offs between out-sorts, precision, recall, fraud avoidance amount, etc.
  • Generates out-sorts & alerts within minutes & incorporates business rules

BACKGROUND

  • 1.2B mobile checks deposited annually worth $480B, 0.4% of checks are fraudulent resulting in returned checks of worth $2B
  • 400% growth in losses reported over 2 years by financial institutions

KEY FEATURES

  • Uses 500+ features to generate out-sorts
  • Captures features from transactions, contact centers, user behavior, device connectivity and geolocation
  • Provides “Pareto Frontier” to show trade-offs between out-sorts, precision, recall, fraud avoidance amount, etc.
  • Generates out-sorts & alerts within minutes & incorporates business rules

BACKGROUND

  • ~$16.8B lost in identify fraud in year 2017, 3.8% increase since 2016
  • ~16.7M consumers were victims of identity theft, 8.5% rise since 2016
  • Account takeover tripled with losses of ~$5.1B, 120% more than 2016

KEY FEATURES

  • Improves detection rate and reduces false positives using Scry’s knowledge graph at various levels such as device, IP & biometrics
  • Assesses real-time risk of user login by aggregating risks from customer 360o profile, fraud patterns and user activity patterns
  • Improves customer experience through white box explanations
  • Reinforcement learning on “its own” and via a “human loop”

BACKGROUND

Conflict of Interest occurs in investment banks & law firms because

  • They cannot represent both parties in merger & acquisition
  • They are often bound by “exclusivity” and related clauses

KEY FEATURES

  • Creates “single source of truth” for entities, i.e., subsidiaries, affiliates, parents
  • Culls out attributes from NDAs and other agreements
  • Identifies news articles with negative news, CXO departures & hostile-takeover
  • Determines “no conflict of interest” situations
  • Provides reasons for conflict of interest & business risks
  • Reinforcement learning on “its own” and via a “human loop”

Key Differentiators & Business Benefits

anomaly-img

Anomaly Detection

Library of 25+ proprietary supervised & unsupervised algorithms, pre-trained & tested for our products

accuracy-img

Knowledge Graphs

30+ probabilistic spatial & temporal graph algorithms to determine links & connected entities

ontology-img

Domain Ontologies

Prebuilt financial ontologies & business rules that each product will improve over time

data-governance-img

Data Cleansing & Harmonization

60+ algorithms for cleansing, harmonizing & validating data from internal & external sources in 30+ formats, e.g., JPEG, video, PDFs

reinforcement-learning-img

External Data Enrichment

40+ scrapers & connectors for ingesting web & paid data subscriptions, e.g., news, social media, traffic, geo-location, sanction lists

Reinforcement-img

Reinforcement Learning

In-built reinforcement learning algorithms that improves our products’ performance over time “with & without a human loop”

api-img

User Interface and APIs

Pre-built graphical user interface (GUI) & APIs for quick deployment & integration with clients’ workflows

deployment-img

Deployment & Scalability

Highly scalable with parallel & distributed computing with options to deploy on-premise or consume as SaaS

business-rules-img

Business Rules

Incorporates clients’ strategies & business rules to improve compliance & accuracy

product-accuracy-img

Product Accuracy

91% – 98% accuracy that improves over time & with high quality ontology

time-cost-img

Time & Cost

60% – 85% improvement in time & cost over the current manual process

feature-engineering-img

Feature Engineering

Incorporates 500+ signal-features & reduce noisy-features to reduce false positives & produce alerts in real-time

For Demo & Additional Information

GET IN TOUCH