Analysis of Financial Transactions for Security – FiT
Financial crime is a scourge whose economic cost is enormous, and which hampers the advent of the digital society. In this context, the fight against fraud, identity theft, or money laundering is a major issue. It relies on the analysis of massive sequences of financial transactions, in order to detect traces of fraudulent activities and take the necessary countermeasures. However, progress in this struggle is hindered by a cruel lack of methods and tools to adequately model and process transactions.
The objective of the labcom FiT is to lead the scientific and technological breakthroughs required for significant progress in detecting anomalies in financial transactions.
The major difficulty of the domain lies in the fact that transaction traces essentially contain information of two natures: the structure of exchanges between accounts and their dynamics over time, as illustrated in the figure below. Most approaches focus on one of the two aspects, whereas the true wealth of data is in the combination of the two. It is this lock that we have to overcome.
Bleckwen is at the forefront of machine learning methods for the detection of fraud. Its solutions are capable of processing transactions in real time and raising alerts for suspicious behavior. It conducts in-house high-level R & D aimed at exploiting increasingly fine features and continuously improving its learning algorithms.
The ComplexNetworks team at LIP6 specializes in network analysis and dynamics. She has developed many of the most used algorithms in the field, and introduced the concept of link flow for modeling temporal interactions. This approach opens up perspectives
unprecedented for integrated analysis of the structure and dynamics of interactions.
Bleckwen and ComplexNetworks want to join forces to model and analyze financial transactions as a flow of links. They will develop and implement the formalisms and algorithms to fully exploit this data. Integrating these advances into Bleckwen products will significantly increase their performance and open new markets for them.
Project coordination
Matthieu Latapy (Sorbonne Université)
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
Partnership
SU Sorbonne Université
Help of the ANR 350,000 euros
Beginning and duration of the scientific project:
September 2019
- 54 Months