AI in Finance Summit schedule

Courtney McGill is the Lead Machine Learning Engineer at Bank of New York Mellon within the Clearance and Collateral Technology (CCT) group.

CCT is responsible for two systemically important systems: Broker Dealer Clearance, which handles the Clearance and Settlement for FED-eligible securities

and Repo-Edge system, which runs the Global Repo Triparty business. Both systems handle multi-trillion USD of assets on daily basis.

Courtney's work is focused on leveraging machine learning and data science to look for opportunities to streamline market efficiency, increase resilience, and reduce overall industry costs.

Courtney holds a Bachelor's degree in Computer Science and Sociology from Colgate University.

Opportunities and Challenges for Automation and AI Adoption

Automation and AI is a key enabler for Credit Suisse to pursue its vision of becoming the most admired bank, which includes establishing exemplary Control, Compliance, and Risk management Processes. Credit Suisse started its process reengineering and automation journey over a couple of years back, with initial focus on RPA. With increasing maturity and deployment of RPA, the focus is now shifting towards more complex automation, to augment human decision making processes, by leveraging Machine Learning and AI. During the session, four key areas will be covered based on experience at Credit Suisse – a) opportunity identification approach across a global bank that has varied business lines (including Swiss Private Bank, Global Markets, International Wealth Management, and Investment Banking & Capital Markets), b) organizational setup (e.g., AI implementation center, data modelling, AI tools management) to drive AI/ automation adoption in a long-term sustainable manner, c) deep dive into select examples (e.g., KYC, Payments processing, asset servicing) of successful deployment of AI/ Automation, and d) challenges and future opportunities/ focus areas (likely to be of interest to systems integrators and startups with product & service offerings in this space)

Uday is leading the process automation and robotics agenda for Credit Suisse, covering the Bank’s various businesses, regions, and shared services functions. In this capacity, he is responsible for opportunity identification through engagement with the various business and functional experts, prioritizing those opportunities to develop viable business cases (with benefits across enhanced controls, risk mitigation, fast time to market, etc.,), ensuring standardization of AI and automation tools used across the bank (so as to not add to the complexity of IT landscape), and driving engagement with AI/ automation vendors (including startups). Prior to Credit Suisse, he spent a few years as a consultant with McKinsey & Co., and was also involved in product development for a couple of Silicon Valley startups. He has Master of Business Administration from Columbia Business School and Bachelor of Engineering from Osmania University, India

AI in Finance: The Current Landscape

Statistical models have always played a big part in financial decision making. I will talk about evolution of statistical modeling in finance and how newer AI methods can complement traditional modeling and enable exploration of previously untapped data. Using concrete real-life examples, I will explain current possibilities and future potential of AI. Where in finance is AI working well? Where is it still evolving? What are the challenges? How do we integrate AI into current business processes? How to build an organization-wide AI strategy? How to build a good AI team? These are some of the questions I will answer based on our experience.

Ambika Sukla leads Morgan Stanley’s AI Center of Excellence, where his team is focused on researching latest developments in AI and applying selective techniques to different areas of finance. His area of expertise ranges from core statistical/probabilistic models to newer deep learning algorithms. The team works on business problems in sales & trading, research, investment management, risk and other divisions. His research focus is on combining newer deep learning models with traditional econometrics, and devising novel unsupervised and semi-supervised models to extract meaning from vast amounts of unlabeled text. Ambika’s background is in signal processing and information theory and he holds a Masters degree in Telecommunication Engineering from NJIT.

Trends for AI in Investments

According to World Economic Forum, 76% of banking CXOs agree that adopting AI will be critical to their organisation’s ability to differentiate in the market. In the meantime, we have seen 48% CAGR in AI investment through 2021; while the global AI investment has exceeded $50 billion influencing more than $1 trillion market up until today. Unlike past 'AI Springs', the science and practice of AI appears poised to continue an unprecedented multi-decade run of advancement. Key financial services opportunities enabled by AI ranging from conservative improvements to bold bets on new capabilities.

Dr Yuanyuan Liu joined AIG in 2013, and is currently leading the machine learning division within AIG’s Investments AI department. During the past 7 years, he has initiated and led multiple global projects such as SME loss-risk analysis, client lifetime value modelling, and opportunity map, etc. Most recently, he is working on AIG’s innovative R&D projects to apply edge-cutting deep learning algorithms in insurance and investment, using generative model, sequential model, and reinforcement learning. Yuanyuan’s team has published a series of papers in NeurIPS, ICML, AAAI, ICASSP for granular and accurate insurance pricing, equity volatility forecasting, efficient multi-mode data samplings etc. Yuanyuan graduated from the University of Oxford with a DPhil in Statistical Machine Learning and a MSc in Applied Statistics. Prior to that, he studied a Mathematics with Statistics major in the University of Bristol.

Peter Brodsky is the CEO of HyperScience, an AI company focused on automating office work. The company delivers automation solutions to Global 2000 companies and government institutions around the world, helping them focus on their core business by reducing bureaucratic burden on them and their customers.

Prior to founding HyperScience, Peter was a Director at Soundcloud, where he led an engineering team that built audio fingerprinting, genre classification and audio recommendations based on audio analysis and user behavior. Peter was also founder and CTO of Instinctiv, a startup that provided powerful software solutions for audio content identification and predictive algorithmic recommendations for media content, and which was acquired by SoundCloud in 2012.

Certifying Robustness to Adversarial Example Attacks in Machine Learning

Adversarial examples that fool machine learning models, particularly deep neural networks, have been a topic of intense research interest, with attacks and defenses being developed in a tight back-and-forth. Most past defenses are best-effort and have been shown to be vulnerable to sophisticated attacks. Recently a set of certified defenses have been introduced, which provide guarantees of robustness to norm-bounded attacks, but they either do not scale to large datasets or are limited in the types of models they can support. This presentation describes the first certified defense that both scales to large networks and datasets (such as Google's Inception network for ImageNet) and applies broadly to arbitrary model types. Our defense is based on a novel connection between robustness against adversarial examples and differential privacy, a cryptographically-inspired technique, that provides a rigorous, generic, and flexible foundation for defense. I discuss the relevance of adversarial example attacks, as well as our defense, for broader applications of machine learning, including the financial sector.

Roxana Geambasu is an Associate Professor of Computer Science at Columbia University and a member of Columbia's Data Sciences Institute. She joined Columbia in Fall 2011 after finishing her Ph.D. at the University of Washington. For her work in cloud and mobile data privacy, she received an Alfred P. Sloan Faculty Fellowship, a Microsoft Research Faculty Fellowship, an NSF CAREER award, a ``Brilliant 10'' Popular Science nomination, an Early Career Award in Cybersecurity from the University of Washington Center for Academic Excellence, the Honorable Mention for the 2013 inaugural Dennis M. Ritchie Doctoral Dissertation Award, a William Chan Dissertation Award, two best paper awards at top systems conferences, and the first Google Ph.D. Fellowship in Cloud Computing.

Deep Learning and Computational Graph Techniques for Derivatives Pricing and Analytics

We review some new approaches from research and literature and Wells Fargo’s work to apply deep learning techniques and computational graph techniques (including algorithmic differentiation) to the solution of high-dimensional forward-backward SDE and PDE in derivative pricing, present some fundamental ideas, applications to derivatives pricing and analytics with some results, and some current and planned extensions

Bernhard Hientzsch is the Head of Model, Library, and Tool Development in the Corporate Model Risk Management Group at Wells Fargo. His group is responsible for the implementation of models, libraries, components, and tools for the validation, benchmarking, and oversight of models at Wells Fargo. Prior to joining Wells Fargo, he was a postdoctoral scientist at New York University in several DoE supported projects and consulting on mathematical, financial, and computer modelling in the USA and Germany. Bernhard received his PhD in applied mathematics from the Courant Institute at New York University.

We suggest a simple practical method to combine the human and artificial intelligence to both learn best investment practices of fund managers, and provide recommendations to improve them. Our approach is based on a combination of Inverse Reinforcement Learning (IRL) and RL. First, the IRL component learns the intent of fund managers as suggested by their trading history, and recovers their implied reward function. At the second step, this reward function is used by a direct RL algorithm to optimize asset allocation decisions. We show that our method is able to improve over the performance of individual fund managers.

Decentralised Algorithms: Harnessing the Power of Collective Predictions

Imagine if anyone could create an algorithm and apply it to vast amounts of data, not just a handful of giant web companies. The Fetch system is based on a concept of Useful Proof of Work, a new consensus mechanism underpinning its distributed ledger that allows large amounts of machine learning computing to occur on a massive decentralised virtual machine. The talk will focus on new approaches to machine learning such as the use of ‘Embeddings’, establishing a trust model for algorithms deployed across the Fetch network as well as examples and approaches for building new ‘bottom-up’ problem solvers whose foundations are rooted in biology.

Toby is Fetch.AI’s Chief Technology Officer and brings with him over a quarter of a century of experience in software architecture and development and more than a decade as CTO across three companies. As a software developer and manager, he’s built and directed several successful computer games including producing the highly successful Creatures series in the 90s. Armed with a fresh, biologically inspired approach to managing software complexity and creating digital intelligence he went on to become Head of Software Design at DeepMind and to construct a shared virtual world architecture called Alice, which powered the world’s most complex online 3D real-time strategy game.

Using Deep Learning with Word Embeddings to improve Customer Satisfaction

Understanding customer satisfaction in retail banking requires exploring and comprehending multiple sources of feedback, such as emails, social networks reviews, web feedback, bot interactions, as well as speech-to-text transcripts collected from call centers. Since such a vast amount of textual data can be difficult to leverage with traditional text mining techniques, deep learning and word embeddings can be used to automatically classify and label feedback, and then deeply analyze and understand their content. In this communication we explain how we leverage all those AI techniques to get an in-depth understanding of the opinions and needs of National Bank’s retail customers. We also show how we improve the performance levels of those AI tools using in-house algorithms and data resources to improve the overall capacity of natural language understanding.

Key Takeaways: • Industrial applications • State of the art classification • Understanding of DL embedding limits

Eric Charton holds a Master in machine learning applied to voice recognition, and a Ph.D. in machine learning applied to Information extraction and natural language generation. He worked as scientist and research project coordinator in academic context in Europe (University of Avignon) and North America (CRIM, École Polytechnique de Montréal) before becoming head of search engine research and development at Yellow Pages Canada. Since March 2018, he is Senior AI Director at National Bank of Canada.

Text Mining Techniques: FINRA & Fraud Detection

Mining and analyzing text helps organizations find valuable business insights in corporate data. Too often we find the common text mining techniques are not effective in many real-world corporate settings, in particular when dealing with short snippets, boilerplate text collected in forms and repeated text due to cut and paste authoring. Choosing the appropriate combinations of the machine learning techniques between supervised (e.g. classification) and unsupervised (e.g. topic modeling or clustering) learning depends on the end goal and the nature of the data (the content, the size and the sparsity, etc.). Even after deciding on clustering approach, there are still various algorithms to consider including well-known K-means and hierarchical agglomerative clustering. In this presentation, I will illustrate a case study and explain how we tackled these challenges.

Yvonne Li joined The New York Stock Exchange (NYSE) Member Regulation Technology as a data architect in 2005 and later as a director at FINRA, the successor to The National Association of Securities Dealers (NASD) and member regulation/enforcement/arbitration of NYSE. For the past 7 years, Yvonne has been working closely with the Office of Risk and Advanced Analytics Team to develop statistical regression and machine learning models for identifying risks at individual, branch and firm levels of broker-dealers by leveraging both structured and unstructured datasets. Prior to joining FINRA, Yvonne was a consultant at JPMorgan Chase for 8 years where she designed and built data warehouse for the Credit Risk and Credit Rating systems. Yvonne also worked at Salomon Brothers supporting the equity trading desk. Yvonne has a BS and MA in Computer Science. Outside of work Yvonne is an avid runner and a passionate saltwater angler.

Chris Merz has been with Mastercard for 18 years. He leads several teams of data scientists that provide an AI layer to such security and decision products as Decision Intelligence, Safety Net, Smart Authentication, AI Express, Stand In, AML, and other R&D efforts. Chris has also spent time in Mastercard Labs, Loyalty Solutions, and Advisors, applying machine learning techniques to engage cardholders. Before joining Mastercard, he worked at a startup for online personalization and performed machine learning research at the NASA COE for AI, and the McDonnell Douglas Research Lab. He has a PhD in Machine Learning from the University of California, a Master of Science in Computer Science from Missouri University of Science and Technology, and a Bachelor of Science in Computer Science from the University of Missouri, St. Louis.

Position Sizing for Pairs Trades

We focus on optimal position sizing in a portfolio of co-integrated assets. We propose a Markov stochastic process optimization model and derive analytical formulae for the dynamic trading strategies. We also conduct an out-of-sample test on historical data in the bitcoin markets. The implementation merely consists of estimating the parameters with a sliding window and the application of the trading strategy, starting in the second window. The algorithm's performance could be enhanced by using reinforcement learning techniques to improve the estimations, and tune the risk aversion parameter which determines our trading volume, with the goal of minimizing transaction costs.

Agnes Tourin grew up in the suburbs of Paris and graduated from the Engineering school MinesParisTech. In 1992, she received her Ph.D in mathematics, from University Paris-Dauphine. Subsequently, she held a tenured position in France, followed by positions at University of Toronto and McMaster University in Ontario. After moving to New York in 2005, and staying at home to care for her two children for 5 years, she got offered her current position at NYU Tandon, where she teaches mathematics for financial engineering. In her research, she applies stochastic control techniques to algorithmic trading.

Investment Selection By Combining Chaos Theory with Artificial Intelligence

Yaron is the CEO of I Know First, an Israeli AI-Wealth Tech company he co-founded in 2011. I Know First specializes on the development and application of advanced self-learning forecasting algorithms for the capital markets, utilizing artificial intelligence and machine learning. Predictions are generated for over 10,000 securities and are applied to discover best investment opportunities, support existing research and investment processes as well as structure AI-powered investment products. The concept of the current algorithm has crystallized following decades of prior research into the nature of chaotic systems by I Know First's R&D team. The algorithm incorporates multi-layered neural networks and genetic algorithms, it is adaptable and scalable allowing comprehensive customized algorithmic solutions depending on clients' needs - banks, asset management first, family offices, hedge funds and self-directed DIY-investors.

After completing his degree in Engineering at the Tel-Aviv University, Yaron worked as a division manager at OIC, one of the Israeli leading consulting firms. Yaron Golgher has over 18 years of experience leading AI, deep learning and machine learning projects including development of AI forecasting algorithms and predictive analytics, quantitative trading strategies, algorithmic trading applications, and big-data solutions for hedge funds and institutional clients.

He received his executive MBA from Ben Gurion University, before founding I Know First.

Investment Selection By Combining Chaos Theory with Artificial Intelligence

The I Know First forecasting algorithm models the markets as non-stationary chaotic systems with fractal properties. Features: Multi-representation – wide and deep supervised learning, data munging to expose important features. Generalization through explicit regularization. Plurality of forecasts. Principal Component Analysis results in near orthogonal inputs. Genetic adaptable algorithm. Can Reinforcement Learning system learn to trade the market? We will report here our results of using Reinforcement Learning to create a trading system based on the I Know First forecasts.
With over 35 years of research in artificial intelligence and machine learning, Dr. Lipa Roitman has developed algorithms to predict chaotic events in financial markets. He has a M.Sc. in chemistry and organic chemistry from Novosibirsk University, a Ph.D. in organic and physical organic chemistry, and photochemistry from the Weizmann Institute of Science, a post doctorate in polymer chemistry from the University of Akron, and has a background in mathematics and statistics. His past work includes: R&D Chemist for Rohm and Haas Corp. USA.

As Co-Founder and CTO of I Know First Ltd., the company uses his algorithm to predict markets and supply big data solutions for institutional and large private investors. Lipa is responsible for the design, development, characterization, specification, and implementation of algorithmic trading solutions. In addition, his other responsibilities include development of quantitative trading strategies, predictive analytics, and development of quantitative trading strategies, algorithmic trading applications, and Big-Data solutions for hedge funds and institutional clients.

AIRMS: A Risk Management Tool Using Machine Learning

We focus on using ML technology as a consultant for trading decisions. In particular, the artificial intelligent risk management system (AIRMS), that is presented, introduces one of the first efforts in the literature to utilize supervised ML as a risk management tool. Two AIRMSs systems are developed based on artificial neural networks and decision trees. These two systems are applied in a FOREX trading breakout strategy classifying its generated signals as profitable and not. Constructing new portfolios using signals classified only as profitable resulted in a significantly enhanced risk profile of the strategy, increasing at the same time the original profit more than 50%.

Spyros Chandrinos grew up in Kalamata, a seaside town in Greece and graduated from the Civil Engineering school of National Technical University of Athens. His thesis results were published in two international scientific journals of Elsevier. He gained experience in the financial sector working in financial institutions in Greece as quantitative trading analyst. Then, he continued his studies in US pursuing a Master's degree in Computational Finance from Carnegie Mellon University. Recently, he worked as a trend following strategies quantitative researcher fellow in KAUST Investment Management Company (one of the largest university endowments in the world with more $25 Billion AUD).

Privacy-preserving Machine Learning in Finance

More (good) data yields better models, but increasing consumer awareness, privacy regulations and proprietary barriers mitigate access to valuable feature sets and our ability to leverage them. The conundrum of computing data without exposing it can be addressed with emerging cryptographic methods such as Secure Multiparty Computation and Fully Homomorphic Encryption. Furthermore, this opens the opportunity to monetize analytics while maintaining data privacy, security, and scarcity value. This talk will discuss the basics of the technologies and various real applications in financial institutions including fraud detection, credit analysis, customer discovery and more.

Dr. Jordan Brandt is the CEO and cofounder at Inpher, a data security company pioneering privacy-preserving machine learning. As a Technology Futurist, Jordan’s research and insight on cybersecurity, AI, robotics and 3D printing have been featured in print and live broadcast internationally on Bloomberg, CNBC, Forbes, Financial Times, Wired and other business and technology press. Jordan is the former CEO and cofounder of Horizontal Systems, acquired by Autodesk (Nasdaq: ADSK) in 2011. He went on to serve as the director of Autodesk’s investment fund, while also teaching and conducting research as a Consulting Professor of Engineering at Stanford University. Jordan completed his undergraduate work at the University of Kansas and his PhD in Building Technology at Harvard. In 2014 he was selected as one of Forbes ‘Next-Gen Innovators’.

Gabrielle Haddad is the Co-Founder & Chief Operating Officer of Sigma Ratings, Inc., the world's first non-credit rating agency. Gabrielle began her career as an M&A attorney at Milbank, Tweed, Hadley & McCloy in NYC. She subsequently spent several years as an executive at The Global Fund, a Geneva, Switzerland based international financing institution, where she worked on risk and governance issues in over 30 country portfolios across Africa, Asia and the Middle East. Her passion for international development, technology and innovation led her to return to the US and to spend a year studying at MIT, where she met her co-founder and launched Sigma. Gabrielle has an undergraduate degree in Finance from Villanova University, a J.D. from The George Washington University Law School and an M.B.A. from the Sloan School of Management at MIT.

Adrien Delle-Case is a Policy Advisor for Digital Finance Regulation & Policy and UniCredit fellow at the Institute of International Finance (IIF) in Washington D.C, the global association of the financial industry. Adrien began his career as Legal Expert in the Compliance department of UniCredit Bank AG (member of the UniCredit Group) in Munich, Germany. He focused on financial crime regulations related to anti-money laundering (AML), financial sanctions and anti-corruption. He subsequently became the Head of UniCredit Bank AG's General Policy Matters AML team and was involved in developing and implementing the bank's AML policies across all business lines, as well as representing the bank in the AML Working Group of the Association of German Banks. Adrien is currently seconded to the IIF's Digital Finance Regulation & Policy team, where he focuses on policy developments in the digital space, with an emphasis on AML and financial data management. Adrien has a law degree from the Ludwig-Maximilians-Universität in Munich and qualified as a lawyer after his clerkship under the responsibility of the Bavarian Ministry of Justice and Consumer Protection.

Data Governance Strategy to Support Machine Learning

For the past decade, enterprise data governance programs have been organized under heavy regulatory scrutiny. Financial services and insurance companies have built up various components of a rigorous data programs largely driven from regulations such as BCBS 239, CCAR, GDPR etc. As a result, many financial companies have built up robust business glossaries and data quality monitoring dashboards and now they seek to gain monetary value from their data. The following session will discuss the fundamentals of data governance as the role of data shifts from compliance towards getting business insights to make informed decisions. At Morgan Stanley Wealth Management, the quality of data is integral to the results of the machine learning models. Incomplete, ill-fit, and unavailable data are just a few scenarios where poor data quality can negatively impact the machine learning results. Data monitoring and proper controls of data are critical components of an enterprise data governance program to ensure that models can be trusted and relied upon for decision-making. We will discuss efforts to provide data transparency to meet regulatory compliance as well as support AI initiatives.

Peggy Tsai is Vice President in the Analytics & Data department at Morgan Stanley Wealth Management. She is responsible for leading the adoption of Data Governance standards and processes across Wealth Management. She works closely with Technology, Business Stewards and Enterprise to improve data quality issues that impact the business. In addition, Peggy works on initiatives to leverage machine learning to improve the capture of data assets across the firm. Peggy has over 15 years of experience in data management, stewardship and governance. Prior to joining Morgan Stanley, Peggy was Data Innovation Lead at AIG where she worked in the enterprise data management practice to support Anti-Money Laundering, Solvency II and GDPR in Latin America and Europe. In addition, she led a winning Innovation Boot Camp team with a proof of concept on Natural Language Processing and Machine Learning aimed at improving the underwriting process. Peggy also worked at S&P Global where she held various managerial positions in enterprise data group and technology in order to drive the value of data between the business and IT. Peggy has a Masters in Information Systems from New York University and a Bachelors of Arts in Economics from Cornell University. She is a member of the Data Governance Professional Organization. In her spare time, she hosts externships for Cornell freshmen students in order to increase the awareness of data and analytics among undergraduates.