Discussion Report 1

Article Summary: Karan Wadhera 

WSJ: U.S., U.K. Collaborate to Spur Innovation in Tech Used to Combat Money-Laundering | June 13, 2022 | Richard VanderfordTop of Form

This article highlights the new program in which U.S. and U.K. collaborate to develop more robust machine-learning technologies to combat money laundering. It will be essential for government and financial institutes to tackle money laundering and other crimes along with maintaining data privacy. The white house and the U.K. government are teaming up on a "prize challenge" program and providing monetary awards for developing a solution to train software to tackle different financial crimes and address problems posed by government agencies.

 

As per government requirements, financial institutes must detect and report any suspicious transactions by their customer. Current rules create massive data, making it difficult for the government and financial institutions to analyze. Governments have encouraged financial institutes for information-sharing partnerships to enhance reporting activities and make data more meaningful. Data privacy rules are one of the challenges in this process. U.S. and U.K. governments want to improve the technology with a new program that will allow machine-learning models to train data from multiple sources without leaving a safe environment—a method known as federated learning.

 

Federated learning technology will ultimately help fight against money laundering and other crimes. This new technology could be a building block for protecting the U.S. financial system from unlawful finance. The prize challenge is expected to open this summer, and the winner will be announced in 2023. FinCEN, the U.K.'s Financial Conduct Authority, and Information Commissioner's Office will make themselves available to innovators as part of the program.

 

This article emphasizes how two countries collaborate to develop new and improved technology to combat financial crimes and money laundering problems. Federated learning technology could be a block to protecting the financial systems from illegal activities. It allows more sharing of financial information between governments and financial institutions and creating robust software for tackling money laundering and other crimes.

 

Bottom of Form

 

SUMMARIZE COMMENTS BELOW

Commenter: Doyin

Top of Form

This article is interesting and it’s my first time learning about Federated Learning in technology. I decided to do some more research and learnt more about how Federated learning works. Federated learning is a machine learning method that enables machine learning models obtain experience from different data sets located in different sites (e.g. local data centers, a central server) without sharing training data. This allows personal data to remain in local sites, reducing possibility of personal data breaches. It does make a lot of sense that financial institutions across multiple countries will be able to use this to combat money laundering through learning models obtained from different data sets located in the separate counties. Federated learning also has some limitations such as:

Data heterogeneity: Models from diverse devices are merged to build a better model in federated learning. Device specific characteristics may limit the generalization of the models from some devices and may reduce accuracy of the next version of the model.

Indirect information leakage: Researchers have considered situations where one of the members of the federation can maliciously attack others by inserting hidden backdoors into the joint global model.

Federated learning is a relatively new machine learning procedure. New studies and research are required to improve its performance

Commenter: Karan

Top of Form

I agree that "Federated learning is a relatively new machine learning procedure. New studies and research are required to improve its performance." In centralized federated learning, the server is responsible for the nodes selection, training process, and aggregation of the received model updates from different nodes, becoming a bottleneck for the system. In decentralized federated learning, nodes can coordinate to obtain the global model. This setup prevents single-point failures as the model updates are exchanged only between interconnected nodes without the orchestration of the central server. Most of the existing Federated learning strategies assume that local models share the same global model architecture in heterogeneous federated learning for heterogeneous clients, e.g., mobile phones and IoT devices. New development in the federated learning framework is called HeteroFL. The HeteroFL technique can enable the training of heterogeneous local models with dynamically varying computation and non-iid data complexities while still producing a single accurate global inference model.

Commenter: Joshua

Top of Form

Karan, 

In a much more subjective inspection of Federated Learning, I too suspect that it will be the future of data privacy for many. Other than the actual objective definition of the computing process, I think it is important to see the big picture of federated learning. It is the capability to securely share data from different places on Earth. Between the U.S. and U.K., this means that we are now able to safely understand suspected and unsuspected data from different perspectives. This might be the most important facit regarding Federated Learning because it is able to link the knowledge of multiple different societies. This means multiple different securities beings used against multiple different attackers. Having the ability to engage a full range of perspectives will undoubtably give the U.S. and other nations the ability to properly detect crimes such as money laundering. 

Commenter: Matthew

One of the most interesting technologies related to finance is data loss prevention (DLP). This is not a new technology, but is coming to a greater prominence as financial leaks seems to be a weekly issue. Some examples of technologies solutions employing DLP would be NetSkope (for the cloud), or at the point of the firewall offered by Palo Alto (not the only provider for DLP for firewalls, but a good example). It is clear that for the aspiring information system analyst, an in-depth understanding of DLP solutions will be an increasingly lucrative area of concern.

ADD A SHORT CONCLUSION

Bottom of Form

INTERNAL USE