it, computing and communications alert (techvision)...convenience warrants better understanding of...

IT, Computing and Communications Alert

(TechVision)

March 18, 2016

D881-TV

Artificial Intelligence Empowering Digital Banking and

Finance Ecosystem

D881-TV 2

Contents

Section Slide Number

Impact of Artificial Intelligence on Banking and Financial Services 3

Artificial Intelligence Impacting Banking and Financial Services–Overview 4

Anti Fraud and Risk Monitor with Deep Learning Neural Networks 5

Machine Learning-based Financial Search Engine 6

Automated Big Data Finance Underwriting Model 7

Management Tool for Structured Financial Products 8

Artificial Intelligence-based Automated Banking Information Platform 9

Wearable Biometric Authentication Solution for Digital Payments 10

Eye-print-based Biometric Authentication Solution for Mobile Payments 11

Facial Recognition-based Digital Payment Solution 12

Strategic Perspectives 13

Industry Contacts 17

D881-TV 3

Impact of Artificial Intelligence on Banking and

Financial Services

D881-TV

Artificial Intelligence Impacting Banking and Financial Service–

Overview

Transformation

• In recent years, banking and finance is experiencing increasing usage of artificial

intelligence (AI) for improved business performance.

• Machine learning, Big Data and data science are all related to building components of AI

systems that feed data from users to the intelligence software. The competition among

financial service providers to reduce cost of services and to maximize customer

convenience warrants better understanding of user preferences and risk management.

Key Benefits

Boosts Productivity

Better

Knowledge

Absorption

Real-time

Feedback

System

Reduces

Cost

Enhanced

Engagement

Increased customized automation through machine learning and

pattern recognition could help financial organizations to offer

more personalized services in real-time at a much reduced cost

as compared to manual process management. Advances in

analytics algorithms and improvements in computational power

are enabling the development of a wide range of service

platforms for the banking and finance sector. Here are some

emerging application areas in banking and financial service

(BFS) that use AI.

Applications Areas

Smart Digital

Wallets

Insurance Decision

Making System

Risk Management

Advanced Biometric

Authentication

Automated

Financial Advisors Fraud Detection

Key Areas

Skymind Deeplearning4j Platform

Anti Fraud and Risk Monitor with Deep Learning Neural Networks Skymind, USA

Why do we need Deep Learning Anti Fraud and Risk Monitors ? On a daily basis, banks have to process millions of transactions, thus their compliance departments have to issue thousands of authorizations per second. During this process, the department is only spared a few milliseconds to analyse and judge whether the transaction is fraudulent or not.

While learning to model fraud, a small sample set from a tiny percentage of total transactions is used to analyse fraudulent materials. This occurrence leads to massive down sampling where 90% of the transactions are not sampled, thus, losing valuable information regarding the evolution of modern frauds.

Finance institutions face an immobile necessity issue where they need to guarantee their users and stakeholders a 99.99% server uptime which is difficult to produce because conventional systems are fragile and are not constantly updated, thus decreasing the reliability of their services.

Which solution can help??

Skymind, an anomaly detection framework for enterprise

service providers, is the supporting backbone for the open-

source framework Deeplearning4j, allowing use of deep

learning into e-commerce, finance and recommended systems.

Key

Technology

Capabilities

The Skymind Deeplearning4j Platform is a huge application enhancement because this innovation will boost the finance industry threat detection facility significantly as this self learning and constantly evolving application will combat fraudulent threats. With the current situation of malicious threats on the Internet, this application will have a great impact on the finance sector, thus increasing its demand in the near future.

Rapid Analysis of

Big Data

Continuous Deep

Learning

Analysis/Results

Cross Platform

Adaptive Framework

Key Benefits from Implementation

Short Term

(1-2 Years)

Medium Term

(3-4 years)

Long Term

(>5 years)

Market Opportunity

High High

Our Thoughts

Medium

The Skymind algorithm is one step ahead of fraudulent threats. The

integrated Deeplearning4j framework enables constant data analysis for

anomalies and also triggers automatic specific responses to disable any

fraudulent related activity.

The Deeplearning4j open-source framework uses the pattern analysing

methodology that is predicted using deep learning that can be used to

monitor patterns for stock-price action, the weather and remote sensor

outputs. The similar learning methodology of patterns runs through the

deep learning framework for high-end analysis using pattern records.

The compact deep learning models that are integrated into the Skymind

Deeplearning4j framework can highly assist low-memory devices, such as

smart mobile devices, tablets and wearables. These devices can understand

and interpret their surroundings, resulting in reduced errors in measured

parameters and data by the device.

AlphaSense Financial Search Engine

Machine Learning-based Financial Search AlphaSense, US

Why do we need Financial Search Engines?

Critical information is scattered all over the Internet in separate sources directed to separate audiences. Information that is necessary for any specific use can be found in multiple sources on the Internet with duplicated data sets rendering the information to be less reliable.

Scattered information causes many ambiguous and misleading situations, where the time used to search for particular information can be dragged to a very long and unproductive time period by the user.

The research process a user undergoes in order to analyse an industry, organization or market is very inefficient and painful as the user has to scheme through big piles of papers by manually going through each research paper using basic skills that is often filled with human errors. Which solution can help??

Key

Technology

Capabilities

Financial news and press releases are constantly being updated on the Internet. The AlphaSense Search Engine is a breakthrough application for the finance industry as this will allow bankers and financiers to immerse themselves in a new finance information portal that will allow its users to gain vast amounts of knowledge related to the finance sector. A positive feedback is expected from the users of this search engine as this innovation brings immense enhancement to the finance industry.

Sophisticated and

Efficient Features

Real time & Smart

Search Results

Large &

Continuously

Updated Database


Short Term (1-2 Years)

Medium Term (3-4 years)

Long Term (>5 years)

Market Opportunity

High High

Our Thoughts

Medium

AlphaSense also has an enhanced Smart Filtering and Alerting framework

that sends powerful email alerts related to the user’s search target and

documents related to their search scope. This framework uses advanced

tagging and relevance filtering functions that operate using machine learning

algorithms to quickly find the most suitable and related content on any topic,

company or industry.

The AlphaSense platforms unique One Search across all documents feature enables

the user to search for a keyword through SEC and global filings, broker research,

conference call transcripts, investor relations presentations, personal in-house content

and real-time news and press releases.

AlphaSense, launched in 2010, is a software technology

startup that serves the financial service industry globally.

AlphaSense applies high-end technology to assist

knowledge and investment professionals to find critical

information related to the finance industry.

The AlphaSense Search Engine is enhanced with the Intelligent Search

framework. This framework combines a number of different sources in a single

location. The framework has features that index every text line intelligently in

order to directly access keywords. The Smart Synonyms technology is

leveraged to enable synonym keyword searches that indefinitely extend the

search criteria based on the search scope.

Enterprise Gamification Platform

Automated Big Data Finance Underwriting Model ZestFinance, USA

Why do we need Financial Underwriting Model? In the Finance industry, traditional lenders have been using the same underwriting methods for the past 40 years. These methods are used to decide whether credit can be offered to the client or not. These methods are based on small assumptions that are vague and often unreliable.

Data on the internet is enormous and is categorized based on cultural, economical or societal context. Machines do not understand the complexity of these hybrid and complex data in a way that humans can naturally observe.

Financial advisors use similar methods to underwrite. These methods are either by using logistic regression, decision trees or combination of both. These methods often lead to incorrect decisions due to the imperfections of the offered models.


ZestFinance, formerly known as ZestCash, consisting of

former Google and Capital One employees, is a team of

mathematicians and computer scientists working on

helping credit lenders of all credit segments with a better

credit risk assessment of new potential borrowers.

Key

Technology

Capabilities

The financial underwriting methodology has been the same for a long time. ZestFinance has taken the initiative to bring a new order of underwriting methodologies in order to provide more trustable and accurate credit decisions. A positive feedback by 2017 is expected to drive this project to achieve its milestones and to provide a stable underwriting model for the financial industry.

Continuous

Learning

Framework

Accurate Analysis High ROI





Market Opportunity

High High

Our Thoughts

Medium

The ZestFinance Model uses its machine learning and neural network

attributes to analyse large amounts of potential credit variables. These

variables range from financial information to technology usage. These

factors are taken into consideration as this will allow a better picture to

assess uncertain factors such as potential for fraud, the risk of default and

the growth of long-term client relationship. The underwriting model

provides high improvement up to 40% by using accurate credit decisions

that lead to increased availability of credit for borrowers and a high return

payment rate for lenders.

The underwriting model by ZestFinance is developed by combining highly

rated data modellers and top tier credit analysts in the finance field to

construct an efficient Big Data model that rapidly and continuously

improves the quality of underwriting for the finance industry.

Quotip Financial Management Platform

Management Tool for Structured Financial Products Quotip, Zurich

Why do we need Financial Management Tools? Digital technologies enhancing business propositions use many different tools and software that provide great improvisation in business analysis .These tools often require many separate authentication protocols and often cause human errors due to separate complex tools and different software output

Organizations handle their daily reports using multiple supporting tools. These tools allow organizations to compile term sheets and basic prospectus separately which can be a continuous troublesome process that decreases employee productivity.

Bankers and financiers rely on information from websites, portals and newspapers to be updated with latest market or industry news coverage. Most of this information is scattered on the Internet and browsing through separate information mediums is time consuming.


Quotip, a start-up in Zurich, is a financial technology

company that provides its users a comprehensive and

integrated sequel of services in three main areas that

are Find, Quote and Report.

Key

Technology

Capabilities

With the rapid growth in Big Data affecting the Finance industry, this positive venture will allow many opportunities for massive finance related content management. Quotip has developed a platform that enables a user-friendly environment for wealth managers to find, price, settle and monitor structured products easily, thus an increase in demand for this application in the finance industry will revolutionize the Financial Products Management services.

Precise Smart Data

Analysis

Automated

Management

Reducing Back-

Office Cost





Market Opportunity

High High

Our Thoughts

Medium

The Quotip Platform is designed with an idea generator framework that is easy

to use. This tool offers complete and spontaneous scenarios of investment

opportunities that are based on predominant market conditions. Modern

machine learning techniques contributed to this framework that allows

indicative pricing capabilities without relying on any sell-side infrastructure.

Quotip used an innovative step to provide its users with full and flexible

overviews for each initiated request by merging the possibility of receiving

manual and automated quotes from within the same workflow. This price

discovery framework assures its users the best execution with a precise

request-for-quote process.

The settlement and clearing technology integrated into Quotip allows massive

reductions in settlement time, mitigation of associated risks and dramatically

reduced back-office cost. On the other hand, the Quotip monitoring and

reporting functionality complies with the latest regulatory requirements of the

financial industry. A customizable notification system integrated into the

platform allows improved client specific experience.

FinGenius Platform

Artificial Intelligence-based Automated Banking Information Platform FinGenius Limited, UK

Why do we need Automated Banking Information Platforms? Banks’ customers find it difficult to communicate with their banks when they require specific information about their finances. Online self-service interfaces are populated with too much information, often failing to answer customers’ specific queries.

Contact center-based assistance is time consuming. Customers are required to provide their details repeatedly to obtain personal financial information, which has a negative impact on customer experience. Customers today demand prompt personalized experience.

New bank employees, on the other hand, find it difficult to obtain domain-specific information about the banks. This makes onboarding processes difficult and time consuming as employees need to go through detailed databases to find the required information.


The FinGenius Platform is an AI-based intelligent and

automated banking information solution that significantly

reduces the complexities of obtaining banking

information for customers and employees.

Key

Technology

Capabilities

The FinGenius Platform seems to be a notable innovation to facilitate self-service banking information access with a personalized experience. The integration of NLP and machine learning empowering the solution with human-like intelligence is a key step forward. The capability of the platform to learn with experience and achieve higher accuracy over time will be a key point of interest for banks to automate information sharing.

High Integration

Potential

Personalized User

Experience

Real-Time Accurate

Information





Market Opportunity

High High

Our Thoughts

Medium

The solution leverages proprietary Natural Language Processing (NLP)

algorithms to understand customer or employee queries in the form of

spoken questions. The algorithms are intelligent enough to understand

different languages and abbreviations and can distinguish between spoken

accents to deliver highly accurate speech recognition in real time.

The Fingenius Platform is powered by proprietary machine learning

algorithms that allow it to perform highly accurate information retrieval.

The capability of these smart algorithms to explore huge volumes of

complex data and deliver contextual outputs with human-like reasoning

gives a human touch to the interaction platform.

The solution has high integration potential and can seamlessly co-exist with

existing infrastructure such as communication channels, customer relationship

management (CRM) and so on. Additionally, the flexibility of the platform to be

deployed either on the cloud or on premise depending on information

sensitivity and requirements ensures optimum security of confidential data.

Nymi Band

Wearable Biometric Authentication Solution for Digital Payments Nymi Inc., Canada

Why do we need Wearable Biometric Authentication for Digital Payments? With the steep decline in the use of payment instruments, such as cheques and cash, digital payment technologies are rendering more secure, flexible and convenient methods for cashless financial transactions.

Assuring security in digital payments has been the biggest challenge faced by solution providers, payment authentication being the most significant one. Incidences of false authentication and abuse of digital payments are growing exponentially.

Traditional card and personal identification number (PIN)-based authentication methods are prone to forgery. Legacy biometric approaches such as fingerprint recognition are also found to be unsafe as they can be easily replicated.


The Nymi Band is a wearable wristband that uses multi-

factor authentication. The solution leverages advanced

AI technology to facilitate contactless secure digital

payments.

Key

Technology

Capabilities

The Nymi Band is a promising innovation, making digital payments more secure. The innovative approach of using continuous heart signals as an authentication modality largely minimizes the complexities involved in providing authentication information every time a transaction is made. The solution, by virtue of its advanced HeartID technology and easy integration capabilities, has the potential to become an authentication standard for contactless payments.

Robust biometric

security

Contactless

payments

High user

convenience





Market Opportunity

High High

Our Thoughts

Medium

The solution consists of an electronics module, namely, Nymi Core. It

leverages an electrocardiogram (ECG) sensor containing two electrodes

that capture ECG data of the user wearing it. The band continues to identify

the user and authenticate payments only till the time it can sense the user

by obtaining the ECG data.

Leverages proprietary HeartID technology to identify individual users

through heartbeat authentication. The technology has been developed

using high-grade proprietary signal processing and machine learning

algorithms for accurately identifying a user through the ECG data

patterns. This unique approach dispenses with the need for repeatedly

collecting biometric data for multiple transactions.

The Nymi Band uses Bluetooth 4.0 Low Energy (BLE) radio to perform

wireless communication with point-of-sale (POS) devices for transactions. The

Nymi Companion Application (NCA), linked with the band and digital payment

cards or accounts, serves as an interface for user enrollment and to facilitate

authentication for payments.

Eyeprint ID

Eye-print-based Biometric Authentication Solution for Mobile Payments EyeVerify Inc., USA

Why do we need Mobile Biometric Authentication? Mobile payments are gaining rapid popularity globally. However, significant security concerns have restrained large-scale adoption of mobile payment solutions. Confidential banking credentials stored on devices are vulnerable to cyber theft.

Usage of personal identification numbers (PINs) or one-time-passwords (OTPs) involves complex procedures and makes mobile payments lengthy. This negatively impacts user experience.

Touch-based mobile biometric solutions, such as on-device fingerprint scanning are often inaccurate. The sensors often become non-responsive due to the accumulation of dirt. Additionally, fingerprints can be easily duplicated to gain unauthorized access.


The Eyeprint ID is an innovative mobile biometric

authentication solution for digital payments. It does not

require any additional expensive embedded sensors and

uses optical biometrics using photographs of eyes..

Key

Technology

Capabilities

The Eyeprint ID solution is a major step forward toward enabling secure biometric authentication for mobile payments even with budget camera phones. The capability of the innovative proprietary backend technology, powered by machine learning, to offer real- time optical biometric authentication, along with FIDO certification, will be key factors toward making it a preferred choice among mobile payment solution providers by 2016.

Real-time

authentication Affordable biometrics

Secure mobile

payments





Market Opportunity

High High

Our Thoughts

Medium

• Eyeprint ID is a software-only optical biometric solution. It leverages a

patented biometric technology that uses proprietary machine learning

algorithms to identify users with respect to patterns of eye veins.

Additionally, the solution uses information from selfie photographs. It

also evaluates the micro features in the eye and its surroundings.

The technology has the capability to transform the eye prints into high

entropy encryption keys to facilitate secure biometric authentication.

The machine learning algorithms allow the solution to identify the

scleral vasculature as well as ocular and periocular micro-features,

which collectively are unique for every human being.

The solution has successfully obtained the Fast IDentity Online (FIDO)

certification which makes it seamlessly interoperable with other innovations.

Additionally, the solution ensures absolute security and reliability of user eye

prints from being stolen as each print is scrambled and encrypted locally on

device. Even if the device is lost or stolen, the prints cannot be abused.

Uniqul

Facial Recognition-based Digital Payment Solution Uniqul Oy, Finland

Why do we need Facial Recognition-based Digital Payment ? Most solutions enabling point-of-sale (POS) digital payment require additional hardware resources, such as cards, tags, smartphones, tablets, and laptops. This makes it essential for the user to carry some devices each time he or she plans to make digital payments.

In the POS digital payments space, most payment solutions focus on conveniently delivering speedy payments. However, security, which is a major concern in any financial transaction, has not been given equal importance, leading to frequent breaches.

The conventional POS digital payment solutions use cards or mobile-based technologies. These solutions are less secure as the chance of theft or abuse of cards or mobile devices exists, which leads to fraudulent payments.


Uniqul is an innovative POS digital payment solution that

leverages facial recognition technology to enable secure

payments without the need for any additional hardware

resources.

Key

Technology

Capabilities

Uniqul’s solution, by virtue of its unique facial recognition technology and cloud-based encrypted payment processing, delivers freedom to the user to make secure, fast, and convenient payments by using just his/her face only. The solution is expected to gain rapid popularity in the world of cashless, cardless, touchless payments and secure digital payments by 2016.

Faster payments High accuracy Cashless, cardless,

touchless payments





Market Opportunity

High High

Our Thoughts

Medium

Uniqul focuses on leveraging biometrics for authentication to authenticate

and enable payment. It uses highly efficient proprietary facial recognition

technology, coupled with proprietary algorithms. The facial recognition

algorithms, based on machine learning, are even capable of distinguishing

between identical twins in real time.

The solution is capable of completing a transaction in less than 5

seconds, in contrast to competing solutions, which usually require more

than 15 seconds to carry out the payment transaction. Additionally, the

military-grade facial-recognition algorithms are efficient enough to

identify users even from a distance, ensuring superfast transactions.

Uniqul’s POS digital payment solution leverages efficient proprietary

encryption techniques developed by vastly experienced in-house security-

technology developers to maintain customers’ financial details, which adds to

the solution’s security aspects. This enhances the reliability, as customers

confidential financial information, linked to the Unique account, stays safe.

D881-TV 13

Strategic Perspectives

14 D881-TV

Key Industry Initiatives

Multi-national banks like Standard Bank, Barclays and Australian Bank ANZ are testing one of the most

advanced artificial intelligence machines–IBM’s Watson–to provide services such as money transfers,

handling customer queries, customer behavior understanding and so on.

The Bank of Tokyo-Mitsubishi has recently employed Nao (a programmable 58cm mini-robot) to

perform reception duties for visitors. The multi-lingual robot deals with requests mainly from foreigners

and offers prerecorded responses for their requests. This robot has been implemented mainly to

improve convenience and increase engagement with customers.

UBS AG, the Switzerland-based financial service company, is working in collaboration with the

Singapore-based pattern detection technology developer Sqreem Technologies Pte. Ltd., to develop a

system that can deliver personalized investment advice to the key clients of the bank.

PayPal, the digital payment handling platform is using deep learning and artificial intelligence

approaches to track its customers activities online. AI helps to improve the cyber security capability of

the platform and enable better fraud detection.

15 D881-TV

Key Patents-World

No. Patent No. Publication Date Title Assignee

1 WO2015036642A1 03-19-2015 Mobile payment system and method based

on a single use token

Pomo Posibilidades S.A.

The invention relates to a mobile payment system and method, said system comprising: a mobile device (4) having an application for generating

tokens according to an item of temporal data that determines the expiration date of the token, a user password and a secret number stored in the

mobile device; and a point-of-sale terminal (5) of a business (1), for receiving the token, user identification data and the amount of the purchase,

and for sending (108) said information to a server (8). The server (8): accesses a user database (6) to validate the token and retrieve bank

information to carry out the payment; accesses a business database (7) to retrieve bank information in order to credit the business (1); and

sends the payment order to a payment gateway (9).

2 WO2008075151A2 06-26-2008 Transaction system and method Fundamo Proprietary Ltd

Rensburg Johannes Janse

Van, Cornilius Johannes

Badenhorst

A transaction system and method is provided in which a plurality of participating system members each has access to a communications device

(2, 3, 4, 5) so as to operatively communicate via an associated network with a computerized server (1 ) in order to instruct the initiation or

conduct of a transaction, typically a financial transaction, by operation of the computerized server consequent on data inputted by a particular

system member. The system includes at least one data base in which there is retained data relating to at least some possible or previous

transactions or both that may be conducted by the particular system member.; Selection means are provided for selecting one of said possible or

previous transactions that best represents data inputted in respect of any particular target transaction on the basis of artificial intelligence such

that data inputted in different ways or with different degrees of accuracy can result in the same target transaction being selected by the selection

means. The selected transaction is communicated to the particular system member whose confirmation is required of the correctness or

otherwise of the selected target transaction.

16 D881-TV

Key Patents–United States

No. Patent No. Publication Date Title Assignee

1 US20100191634A1 07-29-2010 Financial transaction monitoring Bank of America Corp.

Embodiments of the present invention provide systems and methods for monitoring financial transactions. For example, in one embodiment a

system includes a communication interface configured to receive information about each transaction of a plurality of transactions. The system

includes a memory device having a plurality of keywords and an artificial intelligence application stored therein, the plurality of keywords being

associated with a plurality of entities whose transactions are to be specially handled. The system further includes a processor configured to

identify a first group of transactions from the plurality of transactions where the information about each transaction of the first group of

transactions includes at least one of the plurality of keywords. The processor is further configured to then use the artificial intelligence application

to determine that one or more transactions in the first group of transactions are associated with one or more of the plurality of entities.

2 US7319992B2 01-15-2008 Method and apparatus for delivering a virtual reality

environment

Mission Corp.

Described is a method and apparatus for generating a customized dynamic virtual reality environment. The dynamic virtual reality environment

communicates with a participant via a virtual personal assistant utilizing an input/output arrangement. The virtual personal assistant engages the

participant in a natural language conversation to obtain the participant's preferences and personal information. The virtual personal assistant

utilizes an artificial intelligence engine to recognize a plurality of natural languages. Information obtained from the conversation may be translated

into a request for information, services or products. The virtual personal assistant may retrieve any of these from a plurality of remote servers via

a communications network and present the data to the participant. Information obtained from the conversation may be used to evolve both the

virtual reality environment and the virtual personal assistant. Both are continuously learning and adapting to the participant and become more

personalize following every use.

17 D881-TV

Industry Contacts

Ruslan Pisarenko, CEO & CTO , UNIQUL Oy,

Pasilanraitio 9, Helsinki, 00240, Finland.

Phone: +358-45-3144033

E-mail: [email protected]

URL: http://uniqul.com

Jack Kokko CEO, Alphasense,

One Sansome Street, Suite 3500 San Francisco,

CA 94104

Phone: 415-738-8090


URL: https://www.alpha-sense.com

Douglas Clark Merill CEO, ZestFinance,

6636 Hollywood Blvd, Los Angeles, CA 90028

Phone: 323-450-3000


URL: http://www.zestfinance.com/

Dmitry Aksenov, Founder & CEO, FinGenius

Limited,

1 Canada Square, London, UK

Phone: +44-20-7666-3202


URL: www.fingenius.com

Tinna Hung, Director of Marketing, EyeVerify Inc.,

1712 Main Street, 5th Floor, Kansas City, MO 64108

Phone: +1-913-200-2344


URL: www.eyeverify.com

David Buehlmann, Business Development,

Quotip,

Hagenholzstrasse 83b 8050 Zurich, Switzerland

Phone: +41-44-586-30-60

E-mail: [email protected],

URL: http://quotip.com/

Jeffrey Fenton , Marketing Manager, Nymi Inc.,

82 Peter St #500, Toronto, ON M5V 2G5, Canada

Phone: +1-416-977-3042


URL: https://nymi.com/

Adam Gibson CEO, Skymind,

44 Tehama St, San Francisco, CA, 94105.

Phone: +1-406-668-1184


URL: http://www.skymind.io

it, computing and communications alert (techvision)...convenience warrants better understanding of...

Documents