Partners of safeFBDC, the research project of the Financial Big Data Cluster, joined the second Gaia-X hackathon to prototype the first Gaia-X use cases for the financial sector. In collaboration TechQuartier, spotixx, SAP Fioneer and Deloitte from the safeFBDC consortium with support of deltaDAO AG and Ocean Protocol explored Compute-to-Data during the two-day event in early December 2021.
During the Gaia-X hackathon safeFBDC contributed a use case to the “Compute-to-Data & DLT” track and the “Confidential Computing” track. The first use case “AML detection on graphs” aims to advance the development of detection methods for previously undetected cross-institutional patterns in money laundering.
Hackathon use case: “Compute-to-Data & DLT”
safeFBDC intends to iteratively train a global model across many data sources (e.g. Banks). Advanced implementations can entail continuous model aggregation and updates, split learning, randomization, and advanced consensus methodologies.
What is Compute-to-data?
Compute-to-Data is an approach that leaves the data where it is, with the owner, but takes the algorithm to the premise. This way a model can be trained on data of multiple data sources
- without exposing any private data of any data source
- without having a central authority that orchestrates the learning process
- while the data owner keeps full control over their data and over algorithms that are trained on it
- while the full process is transparent and written to an immutable audit trail
Compute-to-Data is especially suitable to remotely train machine learning models on data sets owned by different data providers without being dependent on a central authority. Using Compute-to-Data the data never leaves the premises of the data owners. The data owners remain the only entities in possession of the data. Owners only approve selected and trusted (vested) algorithms to access their data. So, the data owners never lose control over the data.
Minimal Viable Gaia-X Demonstrator — Implementation
Within the “Compute-to-Data & DLT” track of the hackathon deltaDAO and Ocean Protocol provided the Minimal Viable Gaia-X Demonstrator and a first reference implementation of the following Federation Services to facilitate this use case: Gaia-X Portal, Data Contracting Service & Access Controls, Trust & Identity, Data Exchange Logging Service and Federated Catalogue.
The safeFBDC portal acts as frontend to the trusted safeFBDC domain in which three data sets and two algorithms have been integrated to run in the secured Compute-to-Data cluster.
Access controls & data exchange logging services are based on distributed ledger technology (DLT) & smart contracts. Identity & trust services are based on the #opensource ESSIF compliant #SSI Kit by walt.id. The pooling in the safeFBDC domain was enabled by providing a merging algorithm, while the consortium partners remain in strict control of the data at all times. At the heart of the use cases, besides the Gaia-X Federation Services, which are orchestrating and enabling this, is the privacy-preserving “Compute-to-Data” capability of Ocean Protocol. Their distributed ledger automatically creates audit trails and data collection plans that contribute to compliance, data access control and allow provenance of all data assets.
Contributions of the “Confidential Computing” track
The “Confidential Computing” track worked on the provisioning of a trusted execution and storage environment including the key management to create a foundation for the next hackathon and the on-demand creation of necessary infrastructures within the Gaia-X infrastructure, enabled by the Gaia-X Federation Services.
Synthetic Training Data and Algorithms from spotixx
The Frankfurt based startup and TechQuartier-member spotixx provided the necessary synthetic training data and algorithms to perform analytics on the individual and the pooled data. The graph algorithm aims to
- uncover complex networks of criminal actors and organisations,
- track the movement of illicit funds across the entire financial system, traversing multiple persons and institutions
- detect unknown illegal activity that would be impossible to perceive looking at one bank alone
Overall and confidential computing architecture from SAP Fioneer, Deloitte Consulting and TU Darmstadt
SAP Fioneer, Deloitte Consulting and TU Darmstadt collaborate within safeFBDC at the Hackathon to identify new secure and data sovereignty preserving data architectures for safeFBDC Use Cases to exchange data between financial institutes.
Background: What is the safeFBDC?
safeFBDC is a research project funded by the Federal Ministry for Economic Affairs and Energy (BMWi) and orchestrated by TechQuartier. The aim of the project is to enable cross-institutional use of financial market data while preserving individual data sovereignty. It is the use case of Gaia-X in the financial domain and shall serve as a blueprint for data ecosystems made in Europe. For the project time from 2021 to 2023 a consortium, under the leadership of TechQuartier, which is made up of eleven founding and 30+ associated partners covering a large part of the German financial industry and beyond, was formed.
Its goal is to create a unique financial data platform that respects European values and data sovereignty while bringing flexibility, scalability and robustness to state of the art research in Machine Learning and AI to optimize business processes and develop new, data-driven solutions.
Five use cases covering the impact of climate risks on financial markets, the fight against money laundering and improved predictability for monetary policies as well as Stable Supply Chain Finance are researched, developed and prototypically validated using artificial intelligence and machine learning. The interest in the successful realization of safeFBDC is not limited to actors of financial markets; it is intended to serve the greater public good. This is a unique opportunity to establish a strong data ecosystem for the financial services market in Germany.
Consortial leader TechQuartier
As consortial lead of the safeFBDC, TechQuartier supports the evaluation of the demonstrator and the further development of the infrastructure together with the partners into the reference use case for the financial piece in Gaia-X. Many of the startups and partners are already established within the TechQuartier community, whose connections also enable trend scouting and innovation brokerage to this research project.
“Since data is a highly valued commodity of the digital age, we need to ensure the competitiveness of German and European companies and research institutions by giving them safe access to reliable and clean financial data. Currently, this data is being stored in individual financial silos and is not available to the innovators and thinkers who are creating new data-driven solutions for topics like climate risk management and anti-money laundering,”
Dr. Sebastian Schäfer, TechQuartier Managing Director
safeFBDC Partners involved in the Hackathon
With the power of more than 500 startup members fueling cross-industry collaborations, TechQuartier creates possibilities for disruptive technologies, new ideas and modern business models to thrive. Our mission as an innovation broker is to connect and enable our numerous members and stakeholders from industry, academia and government to learn from each other, co-create and succeed in tomorrow‘s digital economy. From our home hub in Frankfurt, Germany, we provide coworking and event space for our curated community of founders and corporate innovators.
spotixx GmbH was founded by a team of financial crime experts in 2019 with a goal to use Artificial Intelligence to improve the detection of criminal activity in the financial system. Today spotixx delivers on this goal by providing machine learning driven solutions that help financial organisations fight against fraud and money laundering.
For fraud prevention spotixx deploys intelligent automated agents either in the cloud or locally within a customers infrastructure to analyse and learn from their transactional data and build human readable rulesets that can be used directly to detect fraudulent transactions. The spotixx agents automatically learn and adapt to new criminal behaviours as they emerge, ensuring that the rulesets for detection are always up to date and effective. For money laundering prevention spotixx analyses network graphs in both single and multi-institutional data to highlight patterns of criminal activity. Spotixx models can be used for direct detection of money laundering or as a complementary tool to other detection paradigms to suggest when likely false positive alerts can be auto-closed, allowing for financial institutions to direct resources towards productive investigations and save costs.
ABOUT SAP FIONEER
Emerged from SAP, a market leader in enterprise application software, SAP Fioneer can build on a heritage of outstanding technology and a thorough understanding of both corporate and consumer demands.
This gives us a head start to bring financial services to the next level with innovative software solutions and platforms. We are helping companies in the financial services industry to meet their needs for speed, scalability, and cost-efficiency through digital business innovation, cloud technology, and solutions that cover banking and insurance processes end-to-end.
ABOUT DELOITTE CONSULTING
Deloitte Consulting is a leader in all areas of digital transformation. The management consultants develop end-to-end solutions from strategy to implementation. Deloitte Consulting uses holistic approaches that are tailored to company and industry-specific conditions, and acts entrepreneurially in the interests of the customer. Deloitte Consulting works in a global network with their customers and leading technology providers. 90% of the Fortune Global 500 companies are Deloitte customers.
The consulting fields include the areas of innovation, strategy, operations (functional as well as industry-specific), technology and human capital. In competence centers such as the Artificial Intelligence Institute, the Blockchain Institute, the Neuroscience Institute, the Digital Ventures and the Digital Factory, strategies and solutions are developed together with customers. Experimental and interactive lab formats are dedicated to central questions about the strategic management of complexity, the initiation of innovative thinking and the generation of added value through data analytics.
Deloitte advises companies in the automotive, banking and insurance, chemical and special materials, energy and raw materials, manufacturing and process industries, retail and consumer goods, health care and life sciences, real estate, media, technology and telecommunications sectors.
ABOUT OCEAN PROTOCOL AND GAIA-X:
Ocean Protocol’s mission is to kickstart a Web3 data economy, giving power back to data owners and enabling people to capture value from data to better our world.
Ocean Protocol already supports a wide range of functionalities that the Gaia-X Federated Services (GXFS) aim to provide. It also works well with Self-Sovereign Identities (SSI) and Verifiable Credentials (VC). It offers a functional, fully open-source foundation to develop Gaia-X towards a fully-fledged data economy in accordance with European values. Ocean Protocol might allow the Gaia-X community and emerging European Data Economy to save years of development and integration time.
Ocean Protocol was named one of 2019’s Hottest Startups by Wired and featured in Harvard Business Review, MIT Technology Review, and Forbes.
deltaDAO AG, located in Hamburg, is the first Ocean Protocol engineering, integration and consulting company, by Ocean Protocol community members. We are working closely with BigchainDB GmbH and the Ocean Protocol Foundation to achieve our common mission to kickstart a European Data Economy. We are a Gaia-X member active in several working groups and initiatives. deltaDAO is a co-organizer of the Gaia-X Hackathon, responsible for the “Compute-to-Data & Distributed Ledger Technology” track where we, as Web 3.0 specialists and integrators, build towards a Minimal Viable Gaia-X with the other Gaia-X community members.
What’s next for safeFBDC and its partners?
The Second Gaia-X hackathon is over and we are happy to share that a second use case including a Federated Learning approach is planned for the near future and next Gaia-X Hackathon.