Ai in financial services

AI in Financial Services
LSE MiM Business Project
1

Prepared for EY
FinTech team
Executive Summary
This report describes research undertaken by The London School of Economics and Political Science on behalf of EY Financial
Services to investigate the use of Artificial Intelligence and Machine Learning and to provide one use case for each of the
following sectors; Insurance, Banking & Capital Markets, and Wealth & Asset Management.
Although there are differing goals for Artificial Intelligence, we understand the ultimate aim of AI to be the pursuit of an
intelligent machine which is able to learn, plan, reason and communicate in natural language. However, we found the industry
has shifted to Applied AI, where the focus lies on solving an isolated problem. Machine Learning is a branch of Applied AI, widely
used and the most prominent type across industries.
The chosen use case in Insurance, Cocoon, illustrates the trend in our research of partnerships between large firms and startup
businesses as Machine Learning coupled with the Internet of Things increasingly becomes a reality.
The chosen use case in Banking and Capital Markets, Kensho, illustrates the trend that technology investments are on the rise. In
fact, Capital Market executives invest more and more of their budget in Machine Learning technologies.
Building an automated investment system that evolves with the market in real-time, Aidyia demonstrates the growing influence
of Machine Learning on Wealth & Asset Management.
2

Prepared for EY
FinTech team
Contents
3
Slide Slide
Introduction
Goals of the presentation
1 Insurance
Banking & Capital Markets
Wealth & Asset Management
Methods
Methodology
Source of Information
6 Discussion
Short/medium/long-term impacts of AI
Conclusions from case studies
36
Taxonomy 8 Recommendations 39
Literature review
AI academic background
Difference in scholarships
Links to management literature
Debate on the impact of AI on management
11 Appendix
• Project Timeline
• Glossary
• Detailed Definitions and Literature Review
• Overview of Industries and Trends
• Use Cases and Tracker
43
Results
Overview of Industries and trends
Criteria for Use Cases
16 Bibliography 95

Prepared for EY
FinTech team4
Lee Baker - Seldon Jason Stockwood – Simply Business
With special thanks for their contributions

Prepared for EY
FinTech team
Goals of the presentation
• The goal of this presentation is to apply academic theory to practice, working
with a real-life business problem to reach quality recommendations.
• This research aims to investigate the use of Artificial Intelligence in Financial
Services and provide one use case for each of the following sectors: Insurance,
Banking & Capital Markets, and Wealth & Asset Management.
• This project also aims to provide an understanding of key definitions in relation
to Artificial Intelligence and Machine Learning, and consider the short, medium
and long term impacts of Machine Learning on Financial Services as a whole.
5

Prepared for EY
FinTech team
Introduction
• At present, the Artificial Intelligence we see within Financial Services and other
industries is applied AI, mainly Machine Learning.
• There is increasing augmentation and automation of manual and cognitive
processes. Applications include fraud detection and chatbots.
• We can already see this in every big bank and major player in the industry, such
as HSBC, Barclays and Zurich Insurance.
• We also see these large players partnering with tech start-ups to develop these
Machine Learning solutions and build them into their own business
propositions.
6

Prepared for EY
FinTech team7
Methods

Prepared for EY
FinTech team
Methodology and Sources
• Interpretative research method – analysis of different methods and
technologies; assessment of their relevance and potential impact
• Qualitative data – Research on definitions and use case applications
• Sources used:
• Computer technology blogs, most current information (2016 only)
• Computer technology encyclopedias
• Industry websites, company websites, blogs
• Academic papers
• Expert interviews (refer to appendix for complete list)
• Consultancy firm reports
• Conferences – Group attended Fintech live!
8

Prepared for EY
FinTech team9
Taxonomy

AI in Financial Services10
John
McCarthy
coins term
First AI
Program “Logic
Theorist”
Negative
Results
for NLP
Turing Test
Light Hill Report
published.
UK Government ceases
AI funding
LISP machines
developed and
marketed
MIT AI
lab
founded
Rise of
desktop
computers
IBMs Deep
Blue beats
chess
champion
Firms invest
over 1
billion in
LISP
Machines
Market for
specialized AI
hardware
collapses
Increasing
computer
power
Focus placed on
solving specific
isolated problems
Major
advances in
AI
Emphasis
on ML
Natural Language used
to make
recommendations
Self Driving
cars
Google
ML
commercially
available on
personal
computers
TimelineAIInterest

Prepared for EY
FinTech team11

Prepared for EY
FinTech team12
Literature Review

Prepared for EY
FinTech team
• 500 BC: Initial interest in humans creating intelligent machines; Greek Mythology
• 1854: ”Laws of thought” introduced by George Boole – introduction of logic
concepts
• 1956: Artificial Intelligence term officially coined by Professor John McCarthy
• 1960-1980: Field branched off in multiple directions and grew rapidly
• 1987-1993: “AI Winter”
13

Prepared for EY
FinTech team
Acting humanly: The Turing Test
approach
- Intelligent behavior: ability to achieve human-
level performance on all cognitive tasks
Thinking humanly: The cognitive
modelling approach
- Need to determine how humans think
- Concern with tracing reasoning steps
Thinking rationally: The laws of
thought approach
- From the field of logic
- Translate problems in logical notation to
create intelligence systems
Acting rationally: The rational agent
approach
- Acting rationally: acting so as to achieve one’s
goals given one’s beliefs
14
More detail in appendix

Prepared for EY
FinTech team
Christenson’s theory of disruption
- Incumbents tend to act on “sustaining innovation”
- Make themselves vulnerable to “disruptive
innovations” at the bottom of the market
- New technology ends up supraceeding existing one
- E.g. Cellular phones disrupting fixed line telephone
- Applying the theory: not necessarily the theory, disruptions occur from the new
technology, evidence of partnership between incumbents and start-ups: why?
15

Prepared for EY
FinTech team
• Oxford University research on 702 occupations: 47% are at risk of being obsolete
in 10-20 years
• Harvard Business Review survey study on 1770 managers in 14 countries: AI will
take over administrative work and leave time for managers to focus on
judgement work, creativity, and social skills
• DeepMind: No evidence that advances in AI are impacting the workforce;
technology is best used to help humans rather than replace them
16

Prepared for EY
FinTech team17
Results

Prepared for EY
FinTech team
Definition of the industries
Insurance is a financial product sold to safeguard individuals, organisations and
their property against the risk of loss, damage or theft.
Banking is the financial dealings of an institution that provides business loans,
credit, savings and checking accounts for companies and for individuals.
Capital markets are markets for the buying and selling financial instruments, such
as equity securities and debt securities.
Wealth management is a high-level professional service that combines financial
and investment advice, accounting and tax services.
Asset management is the sector of the financial services industry that
manages investment funds and segregated client accounts.
18

Prepared for EY
FinTech team
Trends across industries
Insurance
1. AI and the Internet of Things aims to predict risks and improve operational
efficiencies for customers
2. InsurTech: Incumbent partnerships are key
Banking & Capital Markets
1. Responding to customer expectations is the rationale behind technology
transformation
2. Technology investments are on the rise
Wealth & Asset Management
1. Markets are changing as new techniques are being employed
2. Need to embrace AI to sustain competitive advantage
19

Prepared for EY
FinTech team20
Criteria for Use Case
Analysis

Prepared for EY
FinTech team
Criteria Further detail Scale
Impact - Will it disrupt/change in the industry? (+)
- Can it be applied in other industries? (+)
- Are the big players in the industry interested in implementing it? (+)
- Will it create cost savings for the company? (+)
0-4
Feasibility - Is there technology for the idea to be implemented? (+)
- Are there regulatory or ethical roadblocks? (-)
- Is the technology development in a boom phase? (+)
- Data feasibility – availability of data and ease of access to it (+)
0-4
Time - Do we see it starting to appear now?
- Is it a theory in development/ model in progress?
- Is it a technology already in used but underused
- Is it too far in the future/ relying on technological developments?
Short Term: 0 – 3 yrs
Medium Term: 4 – 5 yrs
Long Term: 6 + yrs
21
Criteria

Prepared for EY
FinTech team22
Insurance – deep dive
into use case analysis

Prepared for EY
FinTech team
Proposition Company name Impact Feasibility Final score Time
1. Internet of Things Concirrus 2 2 4 Medium
2. AI Subsound Technology Cocoon 4 4 8 Short
3. Real-time data analytics MetLife Xcelerate 3 2 5 Short
4. Machine learning RiskGenius 2 1 3 Medium
5. Open Source Machine
Learning
Zurich Insurance 2 3 5 Long
6. Automating processes Genworth Financial 3 3 6 Short
Insurance - Overview of use cases
23

Feasibility
Impact
2
4
Concirrus
Cocoon
MetLife Xcelerate
Genworth FinancialZurich Insurance
Insurance – Graph against criteria
24
RiskGenius
Short Term
Long Term
Zurich Insurance
Cocoon
Genworth Financial
MetLife Xcelerate
Concirrus
RiskGenius
Time:
Prepared for EY
FinTech team

Feasibility
Impact
2
4
Concirrus
Cocoon
MetLife Xcelerate
Genworth FinancialZurich Insurance
Insurance – Graph against criteria
25
RiskGenius
Zurich Insurance
Cocoon
Genworth Financial
MetLife Xcelerate
Concirrus
RiskGenius
Time: Short Term
Long Term
Prepared for EY
FinTech team

Short Term Impact
Medium Term Impact
Long Term Impact
Use Case Impact - AI Subsound Technology
Cocoon is an InsurTech start-up that combines
advanced Machine Learning with Internet of Things
in a home security device
• Changes are beginning to
occur as InsurTech
partners with open-
minded incumbents
• As IoT provides the data,
Machine Learning will
then extract the
actionable insight for
insurers
• For customers, IoT
devices will reduce
premiums and improve
their customer service
26
Prepared for EY
FinTech team

Prepared for EY
FinTech team27
Banking & Capital
Markets – deep dive
into use case analysis

Prepared for EY
FinTech team
1. Fraud detection IBM 3 4 7 Short
2. Credit decisioning Logical Glue 2 2 4 Medium
3. Fraud Hub for Gaming Feature Space 4 3 7 Short
4. Financial market predictions Kensho 4 4 8 Short
5. High frequency trading RenTech 3 2 5 Long
Banking & Capital Markets - Overview of
use cases
28

Short Term
Long Term
Feasibility
Impact
2
4
Banking & Capital Markets – Graph against
criteria
29
Time:
Feature Space
Kensho
IBM
Logical Glue
RenTech
Logical Glue RenTech
IBM
Feature
space
Kensho
Prepared for EY
FinTech team

Short Term
Long Term
Feasibility
Impact
2
4
Banking & Capital Markets – Graph against
criteria
30
Time:
Feature Space
Kensho
IBM
Logical Glue
RenTech
IBM
Feature
space
Kensho
Logical Glue RenTech
Prepared for EY
FinTech team

Prepared for EY
FinTech team
Short Term
Medium Term
Long Term
Seeking to replace
equity analysts and thus
generating significant
cost cutting for firms.
Making this technology
more accessible to the
masses.
Creating new business lines in
emerging sectors such as the
commercialisation of space,
autonomous vehicles and
wearable technologies.
Use Case Impact - Kensho
31
Partnered with CNBC, which is running a new
series called #AskKensho
Provides the tools to powerhouse investment
banks to compete with the “quants” that have
taken over the business for the last decade

Prepared for EY
FinTech team32
Wealth & Asset
Management – deep
dive into use case
analysis

Prepared for EY
FinTech team
1. Natural Language Processing Avlien 3 3 6 Short
2. Sentiment Analysis Sensai, Sentifi,
Running Alpha,
Amareos
2 3 5 Short
3. Clusters in real-time AbleMarkets,
AlgoDynamics
2 2 4 Medium
4. Predictive analytics Aidyia, hiHedge,
FNA platform
4 3 7 Medium
Wealth & Asset Management - Overview of
use cases
33

Feasibility
Impact
2
4
Wealth & Asset Management – Graph against
criteria
Time:
Aidyia
Amareos
AlgoDynamix
Aylien
AlgoDynamix
Amareos
Aidyia
Aylien
34
Short Term
Long Term
Prepared for EY
FinTech team

Short Term
Long Term
Feasibility
Impact
2
4
Wealth & Asset Management – Graph against
criteria
Short TermTime:
Aidyia
Amareos
AlgoDynamix
Aylien
AlgoDynamix
Amareos Aylien
Aidyia
35
Prepared for EY
FinTech team

Prepared for EY
FinTech team
Short Term
Medium Term
Long Term
Use Case impact - Aidyia
Decreases risks and costs
for fund managers
Brings differentiated
market position with
unbiased methodology
and better performance
Changes the Asset
Management Industry
completely as the amount
of assets managed by AI
increases
Aidyia has been developing AI-driven strategies
based on deep learning for years. It demonstrates
the ultra application of machine learning and may
change the whole market completely.
36

Prepared for EY
FinTech team37
Discussion

Prepared for EY
FinTech team
Hype Cycle Framework applied to Use Cases
Source:Gartner (July 2016)
38
Prepared for EY
FinTech team

Prepared for EY
FinTech team
Short, Medium and Long-Term Impact
Short-Term Medium-Term Long-Term
• Fraud detection, Anomaly
Detection, Pattern
Recognition, Natural
Language Processing
become the norm
• Rise of RegTech
• Increased demand for data
science talent
• Predictive Analytics –
Tailored customer service
• High Frequency Trading will
be completely automated
• Change management
systems – how to prepare
for an Intelligent Agent
• Displacement of Jobs – even
non-routine and cognitive
roles
• Physical presence on the
high street will become
obsolete
• Restructuring of business
models
39

Prepared for EY
FinTech team40
Summary

Prepared for EY
FinTech team
• Financial institutions may need to consider restructuring their business
models in order to harness this new technology.
• There will be a need to create job positions in relation to AI.
• However, proceed with some caution due to the existing challenges in terms
of trust, privacy and data issues.
• They must also keep an eye on RegTech to better address new regulatory
requirements.
• Overall, firms should embrace AI now or face the possibility of being left
behind.
41
Summary

Prepared for EY
FinTech team42
Thank you.
Any questions?

Prepared for EY
FinTech team43
Stefanie Lieberherr Olivia HotchkissMarta Oliveira Chemsi BennisSophia Wu
The LSE Team

Prepared for EY
FinTech team44
Appendix
• Project Timeline
• Glossary
• Detailed Definitions
• Detailed Literature Review
• Overview of Industries and
Trends
• Insurance Use Cases
• Banking & Capital Markets
Use Cases
• Wealth & Asset
Management Use Cases
• Use Case Tracker

Prepared for EY
FinTech team
Project Timeline
45

Prepared for EY
FinTech team
Project Timeline
46

Prepared for EY
FinTech team
Glossary
47

Prepared for EY
FinTech team
Term Definition
Artificial Intelligence
AI is a subfield of computer science that aims to emulate human intelligence in a machine. The way of achieving
intelligence in a machine and what is meant by intelligence in a machine is where the definitions and goals deviate.
There are two approaches to achieving ‘intelligence’ in a machine, Strong AI and Weak AI.
Strong AI aka. Artificial General Intelligence (AGI) aims to create a machine that can fully perform any action a human
kind.
Weak AI aims to use certain aspects of human reasoning but does not represent the human mind within a machine.
AI Winter
The field of AI research follows the patterns of “hype cycles”, where disappointment and criticism follows the height
of enthusiasm, resulting in funding cuts and then starting the cycle over again with renewed interest years or decades
later.”
1 2
Those periods after the hype experiencing funding cuts are known as the AI Winters.
There have been two prominent AI winters, the first spanning 1974-1980 and the second in 1987-1993.
Artificial Neural Network
Artificial Neural Network (ANN) is an information-processing paradigm that is inspired by the way biological nervous
systems, such as the brain, process information.3
Batch data Analytics
Where data is collected over a period of time, the batch of data is inputted into the system, the data is then
processed, analyzed and an output written
4
Big Data
Very large data sets that need to be analyzed through computational methods to reveal patterns, trends and
behavior.
Binary Logic aka. Boolean
Logic
This is the simplest type of formal logic where there are only ever two values. Usually in the form of ‘True’ or ‘False’
Cognitive Computing The field of cognitive computing is specialised in using algorithms that are based on the way the mind works.
Computational Intelligence
(CI)
CI is a subset of Artificial Intelligence, more specifically weak Artificial intelligence, based on soft computing.
It is often used as a synonym for soft computing.
1
Crevier, Daniel. The Tumultuous History Of The Search For Artificial Intelligence. 1st ed. New York, NY: Basic Books, 1993. Print.
2
"AI Winter". En.wikipedia.org. N.p., 2017. Web. 5 Jan. 2017.
3
Stergiou, Christos and Dimitrios Siganos. "Neural Networks". Doc.ic.ac.uk. N.p., 2017. Web. 21 Jan. 2017.
4
Moise, Izabela, Dirk Helbing, and Evangelos Pournaras. Realtime Data Analytics. Zurich: EZH, 2015. Web. 28 Dec. 2016.
48
Prepared for EY
FinTech team

Prepared for EY
FinTech team
Term Definition
Data Mining aka. Knowledge
Discovery
“It is the process of analyzing data from different perspectives and summarizing it into useful information”
Data Science
“It is an interdisciplinary field about scientific processes and systems to extract knowledge or insights from data in
various forms, either structured or unstructured,
[
which is a continuation of some of the data analysis fields such as
statistics, machine learning, data mining, and predictive analytics”
1
Decision Tree Learning
An algorithm for predictive modelling in machine learning.
2
“This is the most popularly used algorithm for inductive inference which has been successfully applied to a range of
tasks such as learning to assess credit risk of loan applicants. “
Deep Learning
“Deep learning is a sub field and set of algorithms in machine learning that attempt to learn in multiple levels,
corresponding to different levels of abstraction. It typically uses artificial neural networks. The levels in these learned
statistical models correspond to distinct levels of concepts, where higher-level concepts are defined from lower-level
ones, and the same lower-level concepts can help to define many higher-level concepts.
3
These levels of representation and abstraction help to make sense of data such as images, sound, and text.
4
These methods have dramatically improved the state-of-the-art in speech recognition, visual object recognition,
object detection and many other domains such as drug discovery and genomics.”
5
Expert System
“A computer program that represents and reasons with knowledge of some specialist subject with a view to solving
problems or giving advice” it is distinguished from other programs in that it “simulates human reasoning, performs
reasoning over representation of human knowledge and solves problems by heuristic or approximate methods”
6
Expert systems are believed to be among the first successful Artificial Intelligence software.
1
Dhar, Vasant. "Data Science And Prediction". Communications of the ACM 56.12 (2013): 64-73. Web. 21 Jan.2017.
2
Brownlee, Jason. "Classification And Regression Trees For MachineLearning - Machine Learning Mastery". Machine Learning Mastery. N.p., 2016. Web. 23 Dec. 2016.
3
Deng, Li and Dong Yu. "Deep Learning: Methods And Applications".Foundations and Trends® inSignal Processing 7.3-4 (2014): 197-387. Web. 28 Dec. 2016.
4
"Deep Learning Tutorials — Deeplearning 0.1 Documentation". Deeplearning.net. N.p., 2016. Web. 28 Dec. 2016.
5
LeCun, Yann, Yoshua Bengio, and Geoffrey Hinton. "Deep Learning".Nature 521.7553 (2015): 436-444. Web. 27 Dec. 2016.
6 Jackson, Peter. Introduction To Expert Systems. 3rd ed. Addison-Wesley,1998. Print.
49
Prepared for EY
FinTech team

Prepared for EY
FinTech team
Term Definition
Fuzzy Logic
“Is a set of mathematical principles for knowledge representation based on degrees of membership rather than on
crisp membership of classical binary logic.1
“Fuzzy logic provides a way of taking our common-sense knowledge that most things are a matter of degree into
account when a computer is automatically making a decision.”
2
Genetic Algorithms
“A genetic algorithm is a method for solving both constrained and unconstrained optimization problems based on a
natural selection process that mimics biological evolution”3
Hard Computing
(Conventional Computing)
Hard computing or Conventional Computing as it is more commonly known is based on binary logic, meaning two
valued outputs such as yes/no, true/false etc.
Inductive Logic Programming
“Inductive Logic Programming (ILP) is a research area formed at the intersection of Machine Learning and Logic
Programming. ILP systems develop predicate descriptions from examples and background knowledge.”
4
Instance Based Learning
Aka. Memory based learning
“is a family of learning algorithms that, instead of performing explicit generalization, compares new problem instances
with instances seen in training, which have been stored in memory”
5
Internet of Things
“Simply put, this is the concept of basically connecting any device with an on and off switch to the Internet (and/or to
each other). The analyst firm Gartner says that by 2020 there will be over 26 billion connected devices...”
6
LISP Machines These are expert systems that would run the LISP programming
Machine Learning
Machine Learning is a subfield of soft computing and a branch of (weak) Artificial intelligence which "gives computers
the ability to learn without being explicitly programmed (Arthur Samuel 1959)"
7
There are five types of machine
learning; Supervised Learning, Unsupervised Learning, Semi supervised Learning, Deep Learning and Reinforcement
Learning. Machine learning has now become somewhat synonymous with AI, and concerning business applications
data scientists really only refer to machine learning as this is the most prevalent and most applied type of weak AI.
1
Negnevitsky, Michael. Artificial Intelligence. 1st ed. New York: Addison Wesley, 2002. Print.
2
"Glossary - Stottler Henke Associates, Inc.". Stottlerhenke.com. N.p., 2016. Web. 19 Nov. 2016.
3
"Genetic Algorithm". Mathworks.com. N.p., 2016. Web. 3 Jan. 2017.
4
Muggleton, Stephen. "Inductive Logic Programming". N.p., 2016. Web. 27 Dec. 2016.
5
"Instance-Based Learning". En.wikipedia.org. N.p., 2016. Web. 26 Dec. 2016.
6
Morgan, Jacob. "A Simple Explanation Of 'The Internet Of Things'". Forbes.com. N.p., 2014. Web. 3 Jan. 2017.
7
Simon, Phil. Too Big To Ignore. 1st ed. Hoboken, New Jersey: John Wiley & Sons, Inc., 2013. Print.
50
Prepared for EY
FinTech team

Prepared for EY
FinTech team
Term Definition
Natural Language Processing
(NLP)
“NLP applications attempt to understand natural human communication, either written or spoken, and communicate
in return with us using similar, natural language.”1
Real Time data Analytics
“Continually input, process and output data. Data must be processed in a small time period. The analytics ie the
output is always being run and the data input does not stop. This type of processing is important in areas such as
fraud detection.” 2
Reinforcement learning
“Reinforcement learning is a type of machine learning… which allows machines and software agents to automatically
determine the ideal behaviour within a specific context, in order to maximize its performance. Simple reward
feedback is required for the agent to learn its behaviour; this is known as the reinforcement signal.”
3
Representation learning
“Representation learning is a set of methods that allows a machine to be fed with raw data and to automatically
discover the representations needed for detection or classification”
4
Soft Computing
Soft Computing is one of two branches of Weak Artificial Intelligence, the other being Hard Computing. Soft
computing is commonly characterised by the following collection of methodologies: Evolutionary computing, Machine
Learning, Probabilistic Reasoning and Fuzzy Logic.
In contrast to conventional/hard computing, Soft Computing allows for partial truths, imprecision, approximation and
uncertainty.
Supervised Machine Learning
“Supervised learning is where you have input variables (x) and an output variable (Y) and you use an algorithm to
learn the mapping function from the input to the output.”
5
The output data sets are provided.
“The goal is to approximate the mapping function so well that when you have new input data (x) that you can predict
the output variables (Y) for that data.”6 7
1
Marr, Bernard. "What Is The DifferenceBetween Artificial Intelligence And Machine Learning?". Forbes.com. N.p., 2016. Web. 8 Dec. 2016.
2
Moise, Izabela,Dirk Helbing, and Evangelos Pournaras.Realtime Data Analytics. Zurich: EZH,2015.Web. 28Dec. 2016.
3
Champandard, Alex J. "Reinforcement Learning Introduction". Reinforcementlearning.ai-depot.com. N.p., 2016. Web. 27 Dec. 2016.
4
LeCun, Yann, Yoshua Bengio, and Geoffrey Hinton. "Deep Learning".Nature 521.7553 (2015): 436-444. Web. 27 Dec. 2016.
5
6
7
Chapelle, Olivier, Bernhard Schölkopf, and Alexander Zien. Semi-Supervised Learning. 1st ed. Cambridge, Mass.:MIT Press, 2006. Print.
51
Prepared for EY
FinTech team

Prepared for EY
FinTech team
Term Definition
Support Vector Machines
“A Support Vector Machine (SVM) is a discriminative classifier formally defined by a separating hyper plane. In other
words, given labeled training data (supervised learning), the algorithm outputs an optimal hyper plane which
categorizes new examples.”1
Unsupervised Machine
Learning
“Unsupervised learning is where you only have input data (X) and no corresponding output variables.
The goal for unsupervised learning is to model the underlying structure or distribution in the data in order to learn
more about the data.
These are called unsupervised learning because unlike supervised learning there is no correct answers and there is no
teacher. Algorithms are left to their own devises to discover and present the interesting structure in the data.”
2
1
"Introduction To Support Vector Machines — Opencv 2.4.13.2 Documentation".Docs.opencv.org. N.p., 2017. Web. 3 Jan.2017.
2
52
Prepared for EY
FinTech team

Prepared for EY
FinTech team
Detailed Definitions
53

Prepared for EY
FinTech team
The Taxonomy
Source: Mitchell-Guthrie, Polly. "Looking Backwards, Looking Forwards: SAS, Data Mining, And Machine Learning". SAS Blog - Subconscious Musings. N.p., 2014. Web. 17 Dec. 2016.
54

Prepared for EY
FinTech team
Artificial Intelligence is a term, which has to this day found little consensus amongst academics on
its definition. It is a term that has evolved in parallel to the evolution of its application and
implementation which mirrored the evolution of technological capabilities. The primary goal of AI
has however always stayed the same; to create an intelligent machine that would among other
things be able to Learn, Plan, Reason, Communicate in natural language and Represent Knowledge.
What we see being applied to financial services today is more commonly known as Applied AI.
Alan Turing known as the father of AI, and was the first person to invent a test for intelligence in a
machine when he invented the Turing Test in 1950. John McCarthy who successfully set up the first
conference for AI in 1956 coined Artificial Intelligence as a term. By the 1960’s AI research was at its
prime and was heavily funded by the Department of Defence in the United States (US). Other
founders of AI such as Herbert Simon were extremely optimistic at the time about AI’s future stating
"machines will be capable, within twenty years, of doing any work a man can do.”1
55

Prepared for EY
FinTech team
The AI boom was in full swing up until around 1970, and by 1974 the UK and US governments
started cutting funding for AI research due to lack of progress in the field. The years that followed
became known as the first AI Winter. Throughout this first AI winter, research about AI had to be
published under a different name to receive funding. Terms that emerged at the time were pattern
recognition, knowledge based systems, informatics, cognitive systems, computational intelligence
and even machine learning.
The 1980’s saw the end of the first AI winter with the rise of expert systems. However the collapse
of the billion dollar LISP machine industry sunk the AI field into another AI winter at the end of the
1980’s.
The early 21st Century then saw another boom in AI due to increasing computational power and
more focus on solving specific problems.
Advanced statistical techniques (loosely known as deep learning), access to large amounts of data
and faster computers enabled advances in machine learning and perception. By 2010 machine
learning was widely applied all over the world.2
56

Prepared for EY
FinTech team
Soft Computing:
Soft Computing is one of two branches of Weak Artificial Intelligence, the other being Hard
Computing. Soft computing is commonly characterised by the following collection of
methodologies: Evolutionary computing, Machine Learning, Probabilistic Reasoning and Fuzzy Logic.
In contrast to conventional/hard computing, Soft Computing allows for partial truths, imprecision,
approximation and uncertainty.
Machine Learning:
The term was coined by Arthur Samuel in 1959 and is a subfield of soft computing – a branch of
(weak) Artificial intelligence – which "gives computers the ability to learn without being explicitly
programmed” (Arthur Samuel 1959) There are five types of machine learning; Supervised Learning,
Unsupervised Learning, Semi supervised Learning, Deep Learning and Reinforcement Learning
57
Further definitions

Prepared for EY
FinTech team
Applications of ML / where is ML today
• Data Security
• Personal Security
• Financial Trading
• Healthcare
• Marketing Personalization
• Fraud Detection
• Recommendations
• Online searches
• Natural Language Processing
(NLP)
• Smart Cars/Self Driving Cars
Source: Marr, Bernard. "Forbes Welcome". Forbes.com. N.p., 2017. Web. 6 Jan. 2017.
58

Prepared for EY
FinTech team
Detailed Literature Review
59

Prepared for EY
FinTech team
• Interest by humans in creating intelligent machines traces back all the way from
Greek mythology
• Pygmalion myth: A statue brought to life for the love of her sculptor
• 1800s: ”Laws of thought” introduced by George Boole – introduction of logic
concepts
• Artificial intelligence term officially coined by Professor John McCarthy in 1956
• Field branched off in multiple directions, and grew rapidly in the 60s and 70s
• Mid 1980s: “AI Winter”
60

Prepared for EY
FinTech team
Acting humanly: The Turing Test approach
- Intelligent behavior defined as the ability to
achieve human-level performance on all cognitive
tasks
- Capabilities necessary: natural language
processing, knowledge representation, automated
reasoning and machine learning
Thinking humanly: The cognitive modelling
approach
- Need to determine how humans think– through
introspection or psychological experiments
- Concern with tracing reasoning steps and
comparing with human subjects
- Field of cognitive science combining computer
models and psychology
Thinking rationally: The laws of thought
approach
- From the field of logic (first introduced through
Aristotle and his attempt to codify “right thinking”)
- Builds on programs that translate problems in
logical notation to create intelligence systems
- Obstacles: Not easy to state informal knowledge in
formal terms; “in principle” vs. in practice
Acting rationally: The rational agent
approach
- Acting rationally: acting so as to achieve one’s goals
given one’s beliefs = AI becomes the study and
construction of rational agents
- Advantages: more general than “laws of thought”
approach (focus on outcome rather than process),
more amenable to scientific development due to
the clear definition of rationality61

Prepared for EY
FinTech team
• Christenson’s theory of disruption
• Incumbents tend to act on “sustaining innovation”
(making their products more sophisticated and complex)
• Make themselves vulnerable to “disruptive innovations” at the bottom of the
market
• E.g. Cellular phones disrupting fixed line telephony
• Theory: disruptor starts with inferior technology, but as it improves it ends up
supraceeding the existing one
• Applying to ML in FS: not necessarily the theory, disruptions occur from the new
technology, incumbents are trying to partner with them to harness these skills and
not be replaced– change may be due to historical examples of these obsolete
technologies, increasing awareness.
62

Prepared for EY
FinTech team
• Oxford University research on 702 occupations: 47% are at risk of being obsolete
in 10-20 years
• Jobs traditionally automated are manual, we are already seeing cognitive and non-
routine ones being replaced = Management (E.g. Uber, Lyft)
• Harvard Business Review survey study on 1770 managers in 14 countries: argue
AI will take over administrative work and leave time for managers to focus on
judgement work, creativity, and social skills
• DeepMind (as reported by Fortune): No evidence that advances in AI are
impacting the workforce; technology is best used to help humans rather than
replace them
• ”Humans remain the ultimate controller of the systems”
63

Prepared for EY
FinTech team
Overview of Industries and Trends
64

Prepared for EY
FinTech team
Definition of the industry and trends
• Insurance is a financial product sold by insurance companies to safeguard individuals, organisations and / or
their property against the risk of loss, damage or theft
• In the mature insurance industry AI is predicted to impact three main areas:
• There are two key trends highlighted in the literature:
1. AI and the Internet of Things is moving Insurance towards predicting risks before they occur, reducing
claims and improving operational efficiencies for customers
2. Incumbents are partnering with InsurTech with the biggest threats coming from open-minded firms
who are embracing smart startups
Efficiency Competitiveness Risk assessment
65

Prepared for EY
FinTech team
• Banking is the financial dealings of an institution that provides business loans, credit, savings and checking
accounts for companies and for individuals.
• Capital markets are markets for the buying and selling financial instruments. These are equity securities,
which are often known as stocks, and debt securities, which are often known as bonds. Capital markets
involve the issuing of stocks and bonds .
• There are three key trends highlighted in the literature:
1. Nearly ¾ of capital market executives invest more than 11% of their capital budget in technology.
2. 33% say lack of technology is a main obstacle to business transformation in their organization.
3. Nearly 70% of hedge fund traders now use algorithms for 40% of their trading
66
Minimise risk
Increase
efficiency
Reduce cost of
workforce
Sustain
competitive
advantage

Prepared for EY
FinTech team
• Wealth management is a high-level professional service that combines financial and investment advice,
accounting and tax services, retirement planning and legal or estate planning for one set fee.
• Asset management is the sector of the financial services industry that manages investment
funds and segregated client accounts.
• In the mature insurance industry AI is predicted to impact three main areas:
• There are two key trends highlighted in the literature:
• Market are changing as techniques being employed: Techniques, such as natural language processing,
sentiment analysis, clusters in real time and predictive analytics, start to make a difference
• The need to embrace AI: The markets are getting complex and more sophisticated, and tools with
better robustness are needed
67
Reduce Costs
Increase
Performance
Improve Market
Efficiency

Prepared for EY
FinTech team
Insurance Use Cases
68

Prepared for EY
FinTech team
1. Internet of Things: Concirrus
• A platform that is a series of digital insurance underwriting tools that reduce risk
• Concirrus was founded in 2012 to enable businesses to take advantage of connected technologies
and the Internet of Things
• For example, the platform collates and analyses multiple streams of shipping data, giving marine
insurers new insight into both cargo and hull and their associated risks. This intelligence allows
commercial insurers to adjust their risk portfolio and minimise claims costs as a result
Impact IoT has the potential to disrupt the industry and others, however this application has
not been taken up widely yet
Time We are starting to see IoT appear in the industry but further development is needed
Feasibility The technology has yet to reach a boom phase however data should not be a large
roadblock with these applications
69

Prepared for EY
FinTech team
2. AI Subsound Technology: Cocoon
• An InsurTech start-up that combines Machine Learning with Internet of Things
• Cocoon uses unique Subsound® technology to listen for infrasound - subtle, inaudible vibrations
in the air caused by movement – which is then linked to an app that monitors the real-time alerts
• Using advanced Machine Learning, Cocoon is continually learning the unique sound signature of a
home. Out of the box it resembles a blank brain which then learns in 7 – 10 days the ‘normal’
sounds of your home
Impact Can reduce customer premiums and improve service, building on trends around
customer-centricity. It also reduces risk for the insurer
Time Technology is ready to be implemented in the market and is beginning to do so
Feasibility Implemented now by two large insurers - no data feasibility restrictions or regulatory
roadblocks at present
70

Prepared for EY
FinTech team
3. Real-time data analytics: MetLife
• MetLife Xcelerate is a new product allowing insurance quotes in 2 minutes (instead of 20) for
Home and Auto
• It uses public records and consumer reports to gather information about a household's drivers
and vehicles, reducing the amount that needs to be filled into an application form
• "We’re exploring a lot of our access to unstructured data and we’re using machine learning for
that” - MetLife VP of Enterprise Analytics, Malene Haxholdt
Impact There is technology available and it could introduce efficiencies and cost savings across
the entire industry and others
Time This technology is ready to be implemented now
Feasibility There will be regulatory and data roadblocks when trying to access and use customers’
data in this way
71

Prepared for EY
FinTech team
4. Machine learning: RiskGenius
Impact Efficiency savings and compliance applications of Machine Learning have huge
potential for the industry
Time This is still a small startup organisation
Feasibility There could be data extraction issues and regulation regarding the handling of
customer data
• RiskGenius have written algorithms that can break down and understand an Insurance policy
• This algorithm categorizes and structures the content of the policy documents so that they can be
reviewed easily
• The result of this is increased efficiency of sales agents who can quickly review and understand
gaps or compliance issues in a policy
72

Prepared for EY
FinTech team
5. Open Source Machine Learning: Zurich
• Zurich are harnessing Machine Learning with the aim of getting to a more accurate pricing of risk,
increasing the efficiency of claims processes, catching fraud more often and preventing more and
more losses
• “Advanced analytics is one of the top key investments for Zurich because it’s the key differentiator
for insurance companies going into the next couple of decades.” — Conor Jensen, Analytics
Program Director
Impact It is unlikely to disrupt the industry, however Machine Learning and predictive analytics
are being implemented by the big players
Time The impact of this technology has not been seen yet and is still being developed
internally
Feasibility There is technology available however there could be regulation regarding the data
usage and difficulties extracting it
73

Prepared for EY
FinTech team
6. Automating processes: Genworth
• Genworth Financial has automated the underwriting of long-term care (LTC) and life insurance
applications
• A fuzzy logic rules engine encodes the underwriter guidelines and an evolutionary algorithm
optimizes the engine’s performance
• Relying heavily on Artificial Intelligence techniques, the system has been in production since
December 2002 and in 2004 completely automated the underwriting of 19 percent of the LTC
applications
Impact Automating processes has huge cost saving potential for this industry as well as others
Time There is technology available to implement this now into organisations
Feasibility The technology is entering a boom phase as underwriting is increasingly automated
74

Prepared for EY
FinTech team
Banking & Capital Markets Use
Cases
75

Prepared for EY
FinTech team
1. Fraud detection: IBM Research
• IBM Research is using Machine Learning and stream computing to create virtual “data detectives”
in order to detect financial fraud
• The solution analyses historical transaction data to build a model that can detect fraudulent
patterns. The model is then used to process and analyse a large amount of financial transactions
as they happen in real time, also known as stream computing
Impact This combination of ML and stream computing has the potential to disrupt the industry
and others, however this application has not been taken up widely yet.
Time Already in use by a large U.S bank. Results: 15% increase in fraud detection, a 50%
reduction of false alarms and a 60% increase in total savings.
Feasibility The model is first customised to the client’s data and then updated periodically to
cover new fraud patterns.
76

Prepared for EY
FinTech team
2. Real-time predictive analytics for credit
decisioning: Logical Glue
• Logical Glue is a user-friendly software platform for building and deploying predicting models. It
lets businesses use data to automate decision making and increase profitability
• Provides consumer and commercial credit by correctly assessing applicants’ probability of default
• In 2013, Logical Glue were awarded a SMART grant to develop its platform for the creation and
deployment of superior predictive analytics, making it easier than ever before to go from raw
data to accurate models that can be used in real-time
Impact The platform is designed to predict customer behavior for many types of markets,
particularly financial lending, insurance and marketing.
Time Technology is already being used. Clients include lenders, challenger banks, some of
Europe’s leading insurers, a big four bank and firms in other verticals.
Feasibility Implemented already by lenders and banks- no data feasibility restrictions or
regulatory roadblocks at present
77

Prepared for EY
FinTech team
3. Fraud Hub for Gaming: Feature Space
• ARIC Fraud Hub understands player behaviour during real-time game play, to detect the
anomalies which indicate a potential fraud attack.
• Real-time Machine Learning software system for organisations in financial services, including
retail banks, payment providers and card issuers as well as companies in insurance and gaming
• Featurespace was created by a Cambridge University Professor and his PhD student, Dave Excell,
at the forefront and confluence of two academic fields: Data Science and Computer Science
Impact Resulted in 77% reduction in genuine transactions declined, 54% reduction in
chargebacks and 50% reduction in operation costs
Time Already used in some of the world’s largest banks, insurance companies and gaming
industries such as Betfair, William Hill and KPMG.
Feasibility Featurespace’s ARIC platform is live and deployed. No real obstacles
78

Prepared for EY
FinTech team
4. The financial answer machine: Kensho
• A data analytics and Machine Intelligence company
• Combines latest Big Data and ML techniques to analyse how real-world events affect markets
• Offers up a simple Google-style box where you can pose very complex questions in plain English
• Generates original answers to 65 million questions by analysing relationships among more than
90,000 events with graphs and charts within minutes
Impact Significant opportunities to apply their technology beyond financial services, with
applications in government, retail, healthcare and pharmaceuticals.
Time Already in use by Goldman Sachs and the CIA’s venture capital arm.
Feasibility The technology development is in a boom phase by creating new business lines.
79

Prepared for EY
FinTech team
5. The most successful hedge fund:
Renaissance Technologies
• New York-based hedge fund founded in 1982 by James Simons, an award-winning mathematician
• Specializes in systematic trading using only quantitative models derived from mathematical and
statistical analyses; employs mathematical and statistical models for high frequency trading
• Renaissance is one of the first highly successful hedge funds using quantitative trading – known as
“quant hedge funds”
Impact The hedge fund is famed for offering one of the highest return in the hedge-fund
investment world (70%)
Time The technology is there; however no hedge funds or powerhouse investment banks
have been able to replicate it.
Feasibility This technology have been in use for the last 30 years only by the quants
80

Prepared for EY
FinTech team
Wealth & Asset Management Use
Cases
81

Prepared for EY
FinTech team
1. Natural language processing: Aylien
• Aylien Text Analysis API is a package of Natural Language Processing, Information Retrieval and
ML tools for extracting meaning and insight from textual and visual content with ease.
• Text Analysis API: Flexible API that allows one to build ground-breaking Text Analysis solutions
• News API: Search and source news and content from around the web in real time. Stay ahead of
the curve by using the power of Machine Learning and NLP to understand content at scale while
extracting the data that matters to clients.
Impact Make it possible to for machine to understand the sentences and news. Allows many
techniques and service to come true, for example, sentiment analysis
Time Technology is ready and widely adopted in many scenarios.
Feasibility No real obstacles
82

Prepared for EY
FinTech team
2. Sentiment analysis: Amareos
• A leader in Financial News Intelligence. Combining the unique crowd-sourced sentiment data with
in-depth research, it gives clients the edge by providing them with innovative insight into the
drivers of global markets.
• Uses more than 50,000 global news sources, blogs, forums and social media platforms. More than
2 millions articles are analyzed daily.
• Wide Variety of Sentiment Indicators
Impact Makes it possible for machine to understand the feeling and attitude behind words,
which was very difficult before
Time Technology is ready to be implemented now
Feasibility It may have regulatory issues
83

Prepared for EY
FinTech team
• Provides portfolio risk analytics solutions for asset managers, based on sophisticated ‘deep data’
agent-based algorithms scanning in real-time multiple quantitative primary data sources.
• These algorithms analyse the dynamic behaviour of market participants – i.e. buyers and sellers –
and cluster them based on common feature sets. Noise classification, cluster identification and
behavioural finance theory are part of the unique core capabilities.
3. Clusters in real-time: AlgoDynamix
Impact Identifies patterns and helps investors figure out the problems before it is too late
Time Technology is ready to be implemented
Feasibility Patterns and clusters are not self-explanatory, which requires expertise to do further
analysis
84

Prepared for EY
FinTech team
4. Predictive analytics: Aidyia
Impact AI-based trading strategy. Those built on deep learning and neuro network may grant investors
strong and unique positions in the market, which may change the whole market
Time Many funds are already managed by algorithms. More may adopt in the future
Feasibility Intensive optimisation and maintenance are needed
• AI-driven trading strategies: using cutting edge Artificial General Intelligence (AGI) technology to
identify patterns and predict price movements in global financial markets.
• Result is financial prediction and trading systems with a human-like ability to not only recognise
mathematical patterns in market data, but understand what these mean in a broader context.
• Unbiased (No human emotional bias), Optimised (Constantly learning from the market to
increase profits and lower risks), Affordable (Partially replaces experts)
85

Prepared for EY
FinTech team
Use Case Tracker
86

Use Case Tracker (1)
87
Company Industry Function Expertise Contacts Location Impact details
Impact on
industry
Rainbird
(bought by
Mastercard)
Banking
To create an automated, virtual sales assistant. The
AI salesperson will have the work experience
gleaned from the entire sales team and the
thousands of customer conversations, and predict
exactly which calls might convert to sales.
Predictive
Analytics
http://rainbird.a
i/the-
platform/key-
features/
Worldwide
Rainbird helps you actually model your business
world, resulting in a repository of knowledge which
is both scaleable and reusable.You can add
‘probabilistic’ rules throughout your map, enabling
you to create nuanced models that cope well with
uncertainty.
Low
Concirrus Insurance
IoT applications for Insurance
IoT
https://concirru
s.com
UK
IoT will impact the whole industry but Concirrus
itself may not lead these applications
Medium
Paypal Banking
PayPal uses three types of machine learning
algorithms for risk management: linear, neural
network, and deep learning.
Online
payments
system
Worldwide
Fraud prevention will impact the entire banking
industry. However, Paypal's technology is only
being used internally.
Medium
IBM Banking
By using machine learning and stream computing,
IBM creates virtual "data detectives" to detect
financial fraud. The technology analyzes historical
transaction data to build a model that can detect
fraudulent patterns. This model is then used to
process and analyze a large amount of financial
transactions as they happen in real time.
Computer
hardware
company
https://www.res
earch.ibm.com/f
oiling-financial-
fraud.shtml
Worldwide
Fraud costs the financial industry approximately
$80 billion annually U.S. IBM research can help
companies save billions. A large U.S. bank used
IBM machine learning technologies to analyse
credit card transactions and it resulted in an
increase of 15% in fraud detection, a 50%
reduction of false alarms and and a 60% increase in
total savings.
High
Prepared for EY
FinTech team

88
Impact on
industry
Atom Bank Banking
Machine learning technology from the WDS
Virtual Agent software tool use analytics to give
customers the power to self-serve by getting
immediate answers directly from the app.
App-based
bank
http://www.ato
mbank.co.uk/
United
Kingdom
Machine learning makes banking more personal.
The machine can provide "human" touch and all
customers are equal in the eyes of the machine. Medium
Kensho
Capital
Markets
The Financial Answer Machine: Combines latest
big data and machine learning techniques to
analyse how real-world events affect markets.
Data
analytics and
Machine
learning
company
https://www.ke
nsho.com/#/
USA
Significant cost savings by reducing the number of
traders on the trading floor. This software
computes answers with graphs and charts in just a
few minutes which would have taken days,
probably 40 man-hours, from people who were
making an average of $350,000 to $500,000 a year.
High
Rennaissan
ce
Technology
(RenTech)
Capital
Markets
Most successful hedge fund using quantitative
data. RenTech employs mathematical and
statistical models for high frequency trading and
tries to exploit market inneficiencies when large
transaction takes place.
Investment
management
firm
https://www.re
ntec.com/Home
.action?index=tr
ue
USA
Offering to its clients the highest return in the
hedge-fund investment world - 70%. However,
none of its competitors managed to come up with
the same technology.
High
Aylien
Wealth &
Asset
Management
Text Analysis API, a package of Natural Language
Processing, Information Retrieval and Machine
Learning tools for extracting meaning and insight
from textual and visual content with ease
Natural
language
processing
(NLP)
http://aylien.co
m/
Ireland
Essential tools for analysing natural language and
extract meaning and insight
Medium
Prepared for EY
FinTech team

89
Impact on
industry
Sensai
Wealth &
Asset
Management
Data Analytics (Consulting), Sensai combines artificial
intelligence, open source technologies and a new
business analyst friendly query language to reimagine
the way the enterprise interacts with unstructured data
Sentiment
analysis
http://sens.ai/index.ht
ml
USA
Contributes to financial decision
making based on sentiment analysis
(heatmaps and data visualization
etc.) with a tilt towards analytics
and indicators used in trading
systems and risk monitoring.
Medium
Sentifi
Wealth &
Asset
Management
Crowd-based financial market intelligence, the Sentifi
Engine is based on artificial intelligence, machine
learning and semantic methodologies. It is able to
structure Sentifi Signals from millions of unstructured
data shared by the Sentifi Crowd. It also identifies,
profiles, benchmarks and helps engage financial market
participants in the Sentifi Crowd.
Sentiment
analysis
https://sentifi.com/ Switzerland Medium
Running
Alpha
Wealth &
Asset
Management
Visual summaries of global financial markets, powered
by the next evolution of investor sentiment intelligence
and alpha idea generation; Designed for helping
investors build intelligent portfolios that know where
the global influencers will be turning into next
Sentiment
analysis
https://www.runningalp
ha.com/
Canada Medium
Amareos
Wealth &
Asset
Management
Visual summaries of global financial markets, Combining
its unique crowd-sourced sentiment data with in-depth
research, it gives its clients the edge by providing them
with innovative insight into the drivers of global markets
Sentiment
analysis
https://www.amareos.c
om/
Hong Kong Medium
Prepared for EY
FinTech team

90
Impact on
industry
AbleMarkets
Wealth &
Asset
Management
Real-time trading, utilizes Big Data techniques applied to
market data, social media and news to help identify and
manage real-time risks for investment professionals,
including Execution, Portfolio Management and Risk
Management
Clusters in
real-time
http://www.ablema
rkets.com/
USA
It uses real-time data from
markets, looks for patterns and
searches for clusters of traders
who are bailing out of an
investment.
High
AlgoDynamics
Wealth &
Asset
Management
Portfolio risk analytics solutions, risk analytics engine is
based on sophisticated ‘deep data’ agent-based algorithms
scanning in real-time multiple quantitative primary data
sources. These algorithms analyse the dynamic behaviour of
market participants – i.e. buyers and sellers – and cluster
them based on common feature sets. Noise classification,
cluster identification and behavioural finance theory are
part of the unique core capabilities
Clusters in
real-time
http://www.algody
namix.com/
UK High
Aidyia
Wealth &
Asset
Management
AI-driven investment strategies, cutting edge artificial
general intelligence (AGI) technology to identify patterns
and predict price movements in global financial markets.
AGI is a branch of artificial intelligence aimed at learning
mimicking the human brain’s breadth, depth and generality
of understanding. Applied to financial markets, the result is
financial prediction and trading systems with a human like
ability to not only recognize mathematical patterns in
market data, but to understand what these patterns mean in
a broader context.
Predictive
analytics
http://www.aidyia.c
om
Hong Kong
AI-based trading strategy and
risk analysis platform. Those
built on deep learning and
neuro network may grant
investors strong and
uniquepositions in the market.
High
Prepared for EY
FinTech team

91
Impact on
industry
hiHedge
Wealth & Asset
Management
AI-driven trading strategies, using deep reinforcement
learning, the AI trader constantly learn and generate trading
strategies to advance investment goal.
Predictive
analytics
http://www.hihedg
e.com/
Singapore
AI-based trading strategy and
risk analysis platform. Those
built on deep learning and
neuro network may grant
investors strong and unique
positions in the market.
High
FNA
platfrom
Wealth & Asset
Management
FNA Platform features a real-time graph analytics engine
and an advanced client side dashboard that can be
configured for a wide array of use cases within financial
services.
Predictive
analytics
http://www.fna.fi/p
latform
UK High
Sybenetix
Wealth & Asset
Management
Sybenetix provides multi-award winning Market Surveillance
and Compliance Monitoring software. Sybenetix Compass
dramatically reduces false positives whilst also increasing
the speed of investigation.
Conduct
Monitoring
http://www.sybene
tix.com
UK
By learning each trader’s
personality and being more
precise in flagging up suspicious
trading, its system avoids a lot
of costly false alarms.
High
Cocoon Insurance
Cocoon uses unique Subsound® technology to listen for
infrasound - subtle, inaudible vibrations in the air caused by
movement – as a home security device
AI Subsound
Technology
https://cocoon.life UK High
MetLife Insurance
MetLife Xcelerate is a new product allowing insurance
quotes in 2 minutes (instead of 20) for Home and Auto Real-time
Data Analytics
https://www.metlif
e.co.uk
UK Medium
Prepared for EY
FinTech team

92
Impact on
industry
RiskGenius Insurance
RiskGenius takes an insurance policy they’ve written
algorithms that can break it down and understand an
insurance policy
Machine
Learning
http://riskgenius.
com
UK Medium
Zurich Insurance
Zurich are harnessing Machine Learning with the aim of
getting to a more accurate pricing of risk, increasing the
efficiency of claims processes, catching fraud more often
and preventing more and more losses
Open Source
Machine
Learning
https://www.zuri
ch.co.uk/en/pers
onal
UK Medium
Logical Glue Banking
Logical Glue is a user-friendly software platform for
building and deploying predicting models. It provides
consumer and commercial credit depends by correctly
assessing applicants’ probability of default
Real-time
predictive
analytics
company
http://www.logic
alglue.com/
UK
The platform is designed to predict customer
behavior for many types of markets in
different sectors, particularly financial
lending, insurance and marketing.
Medium
Genworth Insurance
Genworth Financial has automated the underwriting of
long-term care (LTC) and life insurance applications
Automating
processes
https://www.gen
worth.com
UK High
Feature
Space
Gaming
ARIC Fraud Hub understands player behaviour during
real-time game play, to detect the anomalies which
indicate a potential fraud attack. Featurespace’s ARIC
platform is a real-time machine learning software system
for organisations in financial services
Fraud hub for
gaming
https://www.feat
urespace.co.uk
UK
Already used in some of the world’s largest
banks, insurance companies and gaming
industries such as Betfair, William Hill and
KPMG.
High
Prepared for EY
FinTech team

Prepared for EY
FinTech team
List of Interviewees
93

Prepared for EY
FinTech team
List of Interviewees
94
Name Title Company Expertise Contact Information Initial contact Interview Date
Jeff Hawkins CEO Numenta Biological Neural Networks N/A contacted 14/12/16 No response
Dr Edgar Whitley
Associate
Professor
The London School
of Economics
Information Systems
e.a.whitley@lse.ac.uk
N/A Unavailable
Gatsby Computational
Neuroscience Unit
Research
Centre
University College
London
Machine Learning & theoretical
Neuroscience
N/A
N/A N/A
Jason Stockwell CEO Simply Business Insurance
jason.stockwood@simplybusin
ess.co.uk
contacted - no reply
yet
19/01/17
Dr Aysha Chaudhary Professor
University College
London
N/A
N/A
N/A N/A
FinTech Connect Live Conference ExCeL London UK’s largest Fintech event www.fintechconnectlive.com N/A 6/7 December
Frank Gorringe
Government
bond trader
UBS
Swiss global financial services
company
N/A
N/A No response
Gah-Yi Ban Professor
London Business
School
Big Data analytics
gban@london.edu
+44 (0)20 7000 8847
N/A N/A
Paolo Cuomo Professional Instech London Innovation in the Insurance Industry paolo@instech.london contacted 12/12/16 22 December 2016
Sabrina Mcewen Professional Cocoon IoT/ Machine Learning in Insurance sabrina@cocoon.life Contacted – email Unavailable
Eddie Litonjua
Project
manager
University College
London
eduardo.litonjua.14@ucl.ac.uk
contacted 13/12 14/12
Stephen H. Muggleton Professor Imperial College Machine Learning s.muggleton@imperial.ac.uk N/A N/A

Prepared for EY
FinTech team95
Bibliography

Prepared for EY
FinTech team
Bibliography
• A-i.com. (2017). Ai Research - Creating a new form of life - History of AI. [online] Available at: http://www.a-
i.com/show_tree.asp?id=3&level=2&root=1
• Aima.cs.berkeley.edu. (2017). Artificial Intelligence: A Modern Approach. [online] Available at: http://aima.cs.berkeley.edu/
• Aggour, K. and Cheetham, W. ‘Automating the Underwriting of Insurance Applications’ General Electric Global Research, 2005. Available at:
http://www.aaai.org/Papers/AAAI/2005/IAAI05-001.pdf
• Andrew Anderson, ‘Why the insurance industry should turn to artificial intelligence to help stop fraud’ 2013. Available at:
http://www.businesszone.co.uk/community-voice/blogs/andrew-anderson/why-the-insurance-industry-should-turn-to-artificial
• ABLE | MARKETS. (2017). ABLE | MARKETS - Observe market activity no human eye can see. [online] Available at:
http://www.ablemarkets.com/am/
• Aidyia.com. (2017). Aidyia | Investing with Advanced Intelligence. [online] Available at: http://www.aidyia.com/
• AlgoDynamixRAP. (2017). AlgoDynamixRAP. [online] Available at: https://algodynamixrap.myshopify.com/
• Amareos. (2017). Amareos. [online] Available at: http://www.amareos.com
• "Artificial Intelligence (AI) And Fintech — Part 1". Chatbots Magazine. N.p., 2016. Web. 20 Jan. 2017.
• "Artificial Intelligence In Financial Services". PwC. Web. 20 Jan. 2017.
• AYLIEN. (2017). AYLIEN | We Bring Intelligence to the Web. [online] Available at: http://aylien.com/
96

Prepared for EY
FinTech team
Bibliography
• Burton, More. "Inside A Moneymaking Machine Like No Other". Bloomberg.com. N.p., 2016. Web. 27 Dec. 2016.
• "Business Banking". Investopedia. N.p., 2016. Web. 27 Dec. 2016.
• Christensen, Clayton M. (1997), The innovator's dilemma: when new technologies cause great firms to fail, Boston, Massachusetts, USA:
Harvard Business School Press, ISBN 978-0-87584-585-2.
• Cwalinski, Kristin. "What Is Kensho?". CNBC. Web. 27 Dec. 2016.
• Deloitte,. Regtech Is The New Fintech. 2016. Print.
• euromoneythoughtleadership, (2016). GHOSTS IN THE MACHINE: Articial intelligence, risks and regulation in financial markets. [online]
Available at: http://www.euromoneythoughtleadership.com/ghostsinthemachine
• Fintech, D. (2017). AI in Digital Wealth Management: Sniffing out investment opportunities | Bank Innovation. [online] Bankinnovation.net.
Available at: http://bankinnovation.net/2016/03/ai-in-digital-wealth-mgt-sniffing-out-investment-opportunities/
• "Forbes Welcome". Forbes.com. N.p., 2015. Web. 27 Dec. 2016.
• Fortune.com. (2017). [online] Available at: http://fortune.com/2016/06/27/five-hottest-fintechs/
• Gary Richardson, KPMG. 2015. Transforming the insurance sector: How machines will change the game for insurers. Available at:
http://www.kpmg.com/Global/en/IssuesAndInsights/ArticlesPublications/Frontiers-in-Finance/Documents/how-machine-learning-fs.pdf
• Gorsht, Reuven (2015): When Machines Replace Middle Management; Forbes BrandVoice. Available at:
http://www.forbes.com/sites/sap/2015/04/12/when-machines-replace-middle-management/#56d6283e63db
97

Prepared for EY
FinTech team
Bibliography
• Harris, G. (2016). A Survey of Deep Learning Techniques Applied to Trading. [Blog] Available at: http://gregharris.info/a-survey-of-deep-
learning-techniques-applied-to-trading/
• hiHedge, AI trading with machine learning. (2017). hiHedge, AI trading with machine learning. [online] Available at:
http://www.hihedge.com/
• "Homepage – Featurespace". Featurespace. Web. 27 Dec. 2016.
• "IBM Research: Machine Learning For Financial Fraud Detection". Research.ibm.com. N.p., 2016. Web. 27 Dec. 2016.
• Investopedia. (2017). Wealth Management. [online] Available at: http://www.investopedia.com/terms/w/wealthmanagement.asp
• Jean-François Gasc, Accenture, ‘Smart startups are helping insurers wise up to the potential of artificial intelligence’. 1 November 2016.
Available at: http://insuranceblog.accenture.com/smart-start-ups-are-helping-insurers-wise-up-to-the-potential-of-artificial-intelligence
• Knight, W. (2017). Will AI-Powered Hedge Funds Outsmart the Market?. [online] MIT Technology Review. Available at:
https://www.technologyreview.com/s/600695/will-ai-powered-hedge-funds-outsmart-the-market/
• Kolbjornsrud, Vegard, Amico, Richard, Thomas, Robert J. (2016): How Artificial Intelligence will Redefine management; Harvard Business
Review. Available at: https://hbr.org/2016/11/how-artificial-intelligence-will-redefine-management
98

Prepared for EY
FinTech team
Bibliography
• "Kensho". Kensho.com. Web. 27 Dec. 2016.
• "Logical Glue: Online, Innovative Predictive Analytics For Financial Services - DIA Barcelona". DIA Barcelona. N.p., 2016. Web. 27
Dec. 2016.
• "Logical Glue | Real-Time Predictive Analytics". Logical Glue. N.p., 2016. Web. 27 Dec. 2016.
• Metz, C. (2017). The Rise of the Artificially Intelligent Hedge Fund. [online] WIRED. Available at:
https://www.wired.com/2016/01/the-rise-of-the-artificially-intelligent-hedge-fund/
• Michael Reilly, Accenture. ‘It starts with a pebble… The impact of AI and robotics in insurance’ 15 September 2016. Available at:
http://insuranceblog.accenture.com/it-starts-with-a-pebble-the-impact-of-ai-and-robotics-in-insurance
• Nadler, Daniel. Interview With Daniel Nadler. in person
• Nanalyze. (2017). Machine Learning For Stock Trading Strategies - Nanalyze. [online] Available at:
http://www.nanalyze.com/2016/04/machine-learning-for-stock-trading-strategies/
• Oliver Ralph, Financial Times ‘Insurance industry faces daunting list of challenges’, 2016. Available at:
https://www.ft.com/content/86aab6c6-21d9-11e6-9d4d-c11776a5124d
99

Prepared for EY
FinTech team
Bibliography
• Parker Beauchamp, Huffington Post. ‘Artificial Intelligence and the Insurance Industry: What You Need to Know’. 10/27/2016. Available at:
http://www.huffingtonpost.com/parker-beauchamp/artificialintelligence-a_4_b_12661858.html
• People.eecs.berkeley.edu. (2017). Introduction to AI: A Modern Approach. [online] Available at:
https://people.eecs.berkeley.edu/~russell/intro.html
• Popper, Nathaniel. "The Robots Are Coming For Wall Street". Nytimes.com. Web. 27 Dec. 2016.
• PwC, ‘AI in Insurance: Hype or reality?’ March 2016 Available at: https://www.pwc.com/us/en/insurance/publications/assets/pwc-top-
issues-artificial-intelligence.pdf
• PwC,. Capital Markets 2020 Will It Change For Good?. 2015. Print.
• "Renaissance Institutional". Rentec.com. Web. 27 Dec. 2016.
• "Renaissance Said Probed By Senate Panel On Tax Maneuver". Bloomberg.com. N.p., 2014. Web. 27 Dec. 2016.
• Runningalpha.com. (2017). Running Alpha | Investing In A Smarter Future™. [online] Available at: https://www.runningalpha.com/
• Sens.ai. (2017). Sensai. [online] Available at: http://sens.ai/index.html
• Sentifi.com. (2017). Financial market insights based on crowd-intelligence and AI - Sentifi. [online] Available at: https://sentifi.com/
100

Prepared for EY
FinTech team
Bibliography
• Staff, Investopedia. "Capital Markets". Investopedia. N.p., 2016. Web. 27 Dec. 2016.
• "The Rise Of AI And Algorithms In The Financial Services Sector - Raconteur". Raconteur. N.p., 2016. Web. 20 Jan. 2017.
• "Top Financial IT Trends For Capital Markets In 2016 - Fintalk". FinTalk. N.p., 2016. Web. 27 Dec. 2016.
• Trevor Lloyd-Jones ‘Machine Learning and Artificial Intelligence in Insurance’ Available at:
https://www2.deloitte.com/content/dam/Deloitte/lu/Documents/operations/lu-intelligent-automation-business-world.pdf
• Vanian, Jonathan (2016): Google Artificial IntelligenceGuru says A.I. Won’t kill job;, Fortune 500. Available at:
http://fortune.com/2016/10/10/google-artificial-intelligence-deepmind-jobs/
• Zuckerman, Gregory. "Renaissance Technologies: Hedge Fund On A $7 Billion Winning Streak". WSJ. N.p., 2016. Web. 27 Dec. 2016.
• Zurich, ‘Zurich announces partnership with Cocoon’ 2016. Available at: https://www.zurich.co.uk/en/about-us/media-centre/general-
insurance-news/2016/zurich-announces-partnership-with-cocoon
• ‘Zurich Insurance on Open Source Machine Learning’, 2016. Available at: http://aibusiness.org/zurich-insurance-on-open-source-machine-
learning/
101

Ai in financial services

More Related Content

What's hot (20)

Similar to Ai in financial services (20)

More from Seldon (20)

Recently uploaded (20)

Ai in financial services